Pinecone jobs
...jobs and webhooks, while implementing robust error handling, rate limiting, and failover mechanisms to ensure production-grade reliability. The AI Specialist will design a sophisticated Retrieval-Augmented Generation (RAG) pipeline using LangChain and LangGraph, where they will select and fine-tune embedding models, architect optimal chunking strategies, and deploy a vector database (such as Pinecone or Chroma) to store both historical reviews and internal business knowledge—including product catalogs, FAQs, brand guidelines, and past customer interactions. This specialist will build a multi-agent orchestration framework where autonomous agents dynamically decide which tools to invoke: querying the vector store for similar past cases, escalating critical negative feedback ...
... * XGBoost / LightGBM / CatBoost * Strong SQL skills and experience working with large datasets. ### Generative AI Hands-on experience with: * Large Language Models (OpenAI, Claude, Gemini, Llama, etc.) * Multi-agent architectures * RAG systems * Prompt engineering * Tool calling / Function calling * Structured outputs * Context management * Memory architectures * Vector databases such as Pinecone, Qdrant, Weaviate, Milvus, Chroma, or FAISS * Embeddings and semantic search * Hybrid retrieval and reranking Experience with one or more frameworks: * LangGraph * LangChain * CrewAI * Google ADK * Semantic Kernel * AutoGen * LlamaIndex --- ## MLOps & Production Engineering Experience with: * Docker * Kubernetes * CI/CD pipelines * MLflow * Model versioning * Model deploym...
...skills and experience: Strong React and JavaScript/TypeScript. Hands-on experience integrating AI/LLM APIs into real products. Comfortable with backend work (Node.js, REST APIs) and a database (PostgreSQL, MongoDB, or similar). Good understanding of async flows, API keys/security, and rate limiting. Attention to detail and clean, maintainable code. Nice to have: Experience with vector databases (Pinecone, pgvector) or RAG pipelines. Familiarity with Next.js. Prior work with real-time features (WebSockets, streaming). To apply, please include: A short note on your relevant experience (React + AI integration specifically). Links to live projects or a portfolio showing similar work. Your availability and estimated hourly rate. Any questions you have about the project. I'm...
...processing pipeline should expose: • high-accuracy OCR for scanned pages, • field-level data extraction that can be fine-tuned per document type, and • a clean JSON export endpoint so downstream systems in law firms or finance teams can consume the results automatically. Expected flow 1. User uploads PDF(s). 2. Service performs OCR where needed. 3. Embeddings are generated and stored (FAISS, Pinecone, or similar—your choice as long as latency stays under two seconds for typical 100-page bundles). 4. A web chat box accepts text input and streams answers back, citing sources. 5. An API call returns the full extracted data set in JSON. Acceptance criteria • Chat responses ≤2 s on 100-page test set. • OCR accuracy on supplied samples ...
...low-quality scanned/PDF inputs, and building pipelines that degrade gracefully rather than silently producing wrong answers. **LLM/RAG** - Production experience building tool-calling LLM architectures with hard output constraints — not prototype chatbots. You should already understand why prompt-level suppression is not a security boundary. - RAG pipeline experience: embeddings, vector DB (ChromaDB, Pinecone, Weaviate, or equivalent), retrieval tuning. - Comfort building schema-enforced, denylist-constrained generation for a regulated domain. **Otto-specific** - Context isolation between subscription tiers within a single LLM session — the candidate must be able to explain, unprompted, why passing full paid-tier data into a shared context window and relying on promp...
...Predictive analysis that flags potential bottlenecks or capacity issues at least two weeks in advance, using past trends. 3. A simple chat interface (web or Slack) so team members can ask follow-up questions and request custom slices of the data. Tech preferences Claude (latest model via Anthropic API) as the core LLM, Python for the orchestration layer, and a lightweight vector store (e.g., Pinecone or Chroma) for context retrieval. I’m open to alternatives if you can justify gains in speed, cost, or accuracy. Deliverables • Clean, well-commented code repository • Deployment scripts (Docker or serverless) • A short read-me explaining how to add new data sources and adjust reporting schedules • One live demo session to walk through setu...
...Sonnet 3.5 deployed in my AWS account (EC2 or an equivalent managed service is fine) and wired up to those embeddings so that users can chat in plain text and receive interactive language-learning guidance based on the content. Here is the workflow I have in mind: • Generate 384-dimensional embeddings for the entire document, verify their quality, and load them into a persistent vector store (Pinecone, Amazon Kendra, or Faiss—whichever you prefer and can justify). • Spin up Claude Sonnet 3.5 in AWS and expose it through a simple web front end or an API endpoint; no voice features are required, text chat only. • Connect the model to the vector store so that retrieval-augmented generation powers the responses. The chatbot’s role is strictly Interact...
...cost-effectiveness, scalability, and clean system architecture. Preferred Technology Stack We welcome equivalent recommendations; however, our preferred technology stack includes: • Frontend: React or • Backend: Python (FastAPI) • AI: OpenAI API (preferred), with support for Anthropic Claude or equivalent • AI Workflow: LangChain or LlamaIndex • Database: PostgreSQL with a Vector Database (e.g., Pinecone, Weaviate, or pgvector) • Cloud: Microsoft Azure or AWS • Authentication: Auth0, Clerk, or equivalent MVP Scope A detailed functional specification, technical requirements, and product documentation will be provided to the selected developer following the execution of a Non-Disclosure Agreement (NDA). What We're Looking For We are seeking ...
...reason through processes, generate outputs, execute actions, communicate with humans, and escalate when human review is required. **Technical architecture we need** We are looking for developers with experience in building advanced AI agent systems using tools such as: * n8n * OpenAI, Claude, Gemini, or other LLMs * RAG systems * Vector databases * Structured databases * Airtable * Supabase * Pinecone * AWS * Azure * APIs * Webhooks * CRM integrations * Gmail and email systems * WhatsApp * Slack * Notion * HubSpot * Holded * Airtable * Lovable * Internal dashboards and client-facing interfaces Our digital workers normally need several layers of reasoning, validation, and execution. Depending on the project, a worker may include 5 to 7 layers, such as: * Input interpretation ...
...logging (Loki), metrics (Prometheus), tracing (Tempo) · Write API docs (Swagger) and deployment runbook · Train the Full Stack team on using the AI APIs --- REQUIREMENTS (Must Have): · Python (5+ years professional) · FastAPI or Flask (expert) · vLLM or TensorRT-LLM (production deployment) · RunPod, AWS SageMaker, or similar GPU cloud · Docker & Docker Compose (advanced) · Vector DBs (Qdrant, Pinecone, or Weaviate) · PostgreSQL with pgvector · Redis (caching + queue) · Git & CI/CD · Linux admin (Ubuntu, shell scripting) Nice to have: Llama/Qwen/DeepSeek experience, AWQ/GPTQ quantisation, NATS, Kong, Vault, Prometheus/Grafana/Loki/Tempo, MinIO/S3, Whisper, FLUX/Stable Diffusion. --- H...
...underlying code relevant from an interview standpoint (not toy examples). Goal: cover realistic scenarios and architectures from interview standpoint around: 1. Agent Anatomy — (Prompt + Tools + Memory) × LLM * Prompting, CoT/ReAct, structured output, model selection * Tool calling, RAG-as-a-tool, short-/long-term memory 2. Knowledge + Topology * Chunking, embeddings, vector DBs (pgvector/Pinecone) * Hybrid retrieval, RRF, reranking, retrieval evaluation metrics * Single-agent ReAct, planning/reflection, orchestrator-worker, handoffs, MCP/A2A 3. Reliability Envelope * Guardrails, prompt-injection defense, LLM-as-judge evaluation * Caching/cost optimization, observability, Human-in-the-Loop (HITL) Note: The above topics are broad; coverin...
migration from pinecone to json format for the db and other changes discussed via chat.
...LinkedIn-ready format (PDF or images). • Publish the generated carousel directly to my LinkedIn account via API/integration. • Maintain a log of generated topics to avoid duplicate content. • Provide a simple dashboard or management interface to upload new documents and manage the knowledge base. Preferred Skills: • RAG systems (LangChain, LlamaIndex, Haystack, etc.) • Vector databases (Pinecone, Qdrant, Weaviate, ChromaDB) • OpenAI, Claude, or Gemini APIs • LinkedIn API integration • Canva API or automated carousel generation • Workflow automation (n8n, Make, Zapier, Airflow, Python) • Cloud deployment and scheduling Nice to Have: • Experience building AI content automation systems • Experience with LinkedIn g...
...Development Team Senior AI Architect with expertise in designing and deploying scalable AI systems. Required Expertise OpenAI APIs and LLMs Agentic AI & Multi-Agent Systems RAG (Retrieval-Augmented Generation) Vector Databases Conversational Memory Systems Context Management Prompt Engineering Function/Tool Calling AI Workflow Orchestration Experience with LangGraph, LangChain, LlamaIndex, Pinecone, Weaviate, Qdrant, or similar technologies is preferred. Travel-Tech Experience (Major Advantage) The AI must understand: Flight booking workflows Fare rules and restrictions Baggage policies Transit and travel requirements Airport intelligence Alternative airport suggestions Layover validation Date optimization OTA-style booking logic API & Backend Requirements Expected...
...Semantic Kernel, AutoGen, CrewAI, Haystack, or similar. * Good understanding of machine learning, NLP, model evaluation, and data preprocessing. * Experience working with APIs, databases, cloud services, and production software systems. * Ability to analyze model outputs, debug failures, and improve system quality. ## Preferred Qualifications * Experience with vector databases such as Pinecone, Weaviate, Milvus, FAISS, Qdrant, Chroma, or Azure AI Search. * Experience with cloud platforms such as Azure, AWS, or Google Cloud. * Experience with model serving or inference frameworks such as vLLM, TGI, Triton, Ray Serve, or FastAPI. * Experience building production-grade AI agents, chatbots, copilots, document intelligence systems, or workflow automation tools. * Knowledg...
...Semantic Kernel, AutoGen, CrewAI, Haystack, or similar. * Good understanding of machine learning, NLP, model evaluation, and data preprocessing. * Experience working with APIs, databases, cloud services, and production software systems. * Ability to analyze model outputs, debug failures, and improve system quality. ## Preferred Qualifications * Experience with vector databases such as Pinecone, Weaviate, Milvus, FAISS, Qdrant, Chroma, or Azure AI Search. * Experience with cloud platforms such as Azure, AWS, or Google Cloud. * Experience with model serving or inference frameworks such as vLLM, TGI, Triton, Ray Serve, or FastAPI. * Experience building production-grade AI agents, chatbots, copilots, document intelligence systems, or workflow automation tools. * Knowledg...
...Sonnet 3.5 deployed in my AWS account (EC2 or an equivalent managed service is fine) and wired up to those embeddings so that users can chat in plain text and receive interactive language-learning guidance based on the content. Here is the workflow I have in mind: • Generate 384-dimensional embeddings for the entire document, verify their quality, and load them into a persistent vector store (Pinecone, Amazon Kendra, or Faiss—whichever you prefer and can justify). • Spin up Claude Sonnet 3.5 in AWS and expose it through a simple web front end or an API endpoint; no voice features are required, text chat only. • Connect the model to the vector store so that retrieval-augmented generation powers the responses. The chatbot’s role is strictly Interact...
...Sonnet 3.5 deployed in my AWS account (EC2 or an equivalent managed service is fine) and wired up to those embeddings so that users can chat in plain text and receive interactive language-learning guidance based on the content. Here is the workflow I have in mind: • Generate 384-dimensional embeddings for the entire document, verify their quality, and load them into a persistent vector store (Pinecone, Amazon Kendra, or Faiss—whichever you prefer and can justify). • Spin up Claude Sonnet 3.5 in AWS and expose it through a simple web front end or an API endpoint; no voice features are required, text chat only. • Connect the model to the vector store so that retrieval-augmented generation powers the responses. The chatbot’s role is strictly Interact...
...Infrastructure & Security – Deploy on AWS/GCP/Azure with Docker and Kubernetes, implement secure authentication (JWT/OAuth), and ensure data privacy and system resilience. You Have: • Proven expertise building LLM-powered systems and agentic AI applications • Hands-on experience with LangChain, AutoGen, or CrewAI • Strong Python backend development skills • Proficiency with vector databases (FAISS, Pinecone, Weaviate) • Cloud deployment and scalable architecture experience • Familiarity with OpenAI, Anthropic, Hugging Face APIs Nice to Have: Multi-agent collaboration systems, AI evaluation/monitoring, production AI system experience This role combines full-stack development with frontier AI—ideal for engineers passionate about bui...
...mode: detects emergency keywords across all 6 languages and instantly switches to direct, no-emoji, no-slang response mode • WhatsApp channel working end-to-end (Meta Cloud API or Twilio sandbox) • Simple web chat fallback UI if WhatsApp setup takes too long --- TECH STACK EXPECTED • LLM: Claude API (Anthropic) or OpenAI GPT-4o or Equivalent • RAG: pgvector or lightweight vector store (Chroma, Pinecone free tier) • WhatsApp: Meta Cloud API or Twilio sandbox • Backend: Node.js / NestJS or Python FastAPI • Demo KB: we provide PDFs, Word files, and images for at least 10 destinations/operators --- TIMELINE Strict 48-hour delivery from project start. This is a demo, not production — clean code matters but speed is the priority. --- DEL...
...without touching code. – User accounts with basic matter tracking, document uploads and encrypted storage. – Deployment to my existing domains with staging and production environments. Stack guidance I’m comfortable with React or on the front end and Node, Django or similar on the back, so long as you can justify your choice. For AI, feel free to pull in OpenAI / GPT-4, LangChain, Pinecone (or an equivalent vector store). You just need to justify your choice and it's fine if the team is okay with it. Acceptance criteria (final hand-off) 1. Two responsive websites live on our servers, SSL enabled. 2. AI chatbot answers our seeded questions with <3 sec average latency and 95 % accuracy on a test set we’ll supply. 3. Admin dashboard dem...
...should use RAG to answer questions based on the knowledge base. The developer should be familiar with: * Embeddings * Vector databases * Semantic search * Document chunking * Retrieval pipelines * Prompting with retrieved context * Reducing hallucinations * Handling “I don’t know” cases when the answer is not in the knowledge base Possible vector database options include: * Qdrant * Weaviate * Pinecone * Chroma * PostgreSQL with pgvector * Another recommended option The final solution should ensure that the chatbot only answers based on the available knowledge base whenever possible. ## Expected Deliverables The selected freelancer will be expected to deliver: * A working Hexabot-based WhatsApp chatbot. * A configured WhatsApp integration. * A Wiki-based...
...Google Sheets, Airtable or Supabase * Data normalization Deliverables 1. Fix Help Scout attachment extraction 2. Download and process quotation PDFs automatically 3. Extract structured quote information using AI 4. Store extracted data in a database (Google Sheets initially) 5. Document the workflow 6. Provide recommendations for scaling to a vector database (Pinecone or Supabase) Bonus Experience with: * RAG systems * Pinecone * Supabase Vector Search * AI-powered quote recommendation systems End Goal I want an AI assistant that can receive inquiries such as: "Ég þarf 50 boli með logo" (in Icelandic meaning: I need 50 t-shirts) and instantly retrieve similar historical quotations, showing: * closest quantity below * closest quantity ...
Title: AI Multilingual Search Platform Developer Needed (Mobile App + Website) Description: I am l...PAN/tax * company registration * municipality procedures * police procedures * required documents * official processes Technical Requirements: * OpenAI API integration * RAG AI architecture * Vector database integration * OCR for scanned documents * Nepali Unicode support * Scalable backend architecture Preferred Stack: * Flutter * React * Python FastAPI or Node.js * PostgreSQL * pgvector/Pinecone Requirements: * Previous AI/RAG experience * Full source code delivery * Clean scalable architecture * Deployment support * Good UI/UX Please send: * Similar projects completed * Estimated budget * Estimated timeline * Team size * Recommended architecture * Monthly API/server cost...
...chatbots, pre-built integrations (Intercom, HubSpot, Xero) powered by their own knowledge base. We are model-agnostic by design. REQUIRED: BOTH TIERS This role requires genuine production experience across both tiers. Please only apply if you can honestly demonstrate both. TIER 1 — SMB Delivery: · n8n (workflow orchestration) · Notion (knowledge base and 2nd Brain) · OpenRouter (model routing) · Pinecone / Supabase pgvector (vector DB) · RAG pipelines — chunking, embedding, retrieval · Slack bot deployment · DigitalOcean (hosting and snapshot deployments) · Pre-built AI agent integrations (Intercom, HubSpot, etc.) TIER 2 — Enterprise Delivery: · LangChain / LlamaIndex (Python) · Azure Open...
...Sonnet 3.5 deployed in my AWS account (EC2 or an equivalent managed service is fine) and wired up to those embeddings so that users can chat in plain text and receive interactive language-learning guidance based on the content. Here is the workflow I have in mind: • Generate 384-dimensional embeddings for the entire document, verify their quality, and load them into a persistent vector store (Pinecone, Amazon Kendra, or Faiss—whichever you prefer and can justify). • Spin up Claude Sonnet 3.5 in AWS and expose it through a simple web front end or an API endpoint; no voice features are required, text chat only. • Connect the model to the vector store so that retrieval-augmented generation powers the responses. The chatbot’s role is strictly Interact...
...Requirements We require a dashboard/interface for our team to interact with the AI and manage the generated assets. The required feature modules include: 1. Data Ingestion & RAG (Retrieval-Augmented Generation) Ability to upload raw PDFs and course overviews. System must accurately parse text, tables, and basic diagrams to use as context for content generation. Suggested tools: LlamaParse, Pinecone/Milvus, OpenAI models. 2. LLM Orchestration & Script Generation AI generation of highly structured course outlines, lecture scripts, and quiz questions based strictly on the uploaded source materials. Human-in-the-loop interface: The content creator must be able to review, edit, and approve all generated text scripts before they are sent to expensive media APIs (like vid...
...a plain-language instruction from the chat UI, break it into a basic sequence of sub-tasks, and then execute each step without additional prompts. Typical actions include: opening pages through Playwright or Selenium, reading or writing local files, running terminal commands, and calling external REST APIs. Throughout the run it should store and retrieve context via a vector database (Chroma, Pinecone, or FAISS) so conversations remain coherent and previously gathered knowledge is reusable. Architecture guidelines • Language: Python • Framework: FastAPI or Flask (whichever you’re faster with) • Orchestration: LangChain or LangGraph for tool routing and memory • Containerisation: Docker, with an easy “docker compose up” to launch ev...
...never notice a hand-off to a human. Here is what I need you to do: • Design and implement the core large-language-model pipeline (GPT-4, Claude, or another strong model of your choice). • Integrate retrieval-augmented generation so the bot can pull from my existing knowledge base and keep answers grounded. • Orchestrate the prompts, embeddings, and vector search (LangChain, LlamaIndex, Pinecone or similar) for speed and reliability. • Wrap the model in a clean, well-documented API that I can drop into a web or mobile front end. *Cost optimization by combining different models Acceptance criteria – The chat responds in under two seconds for 95 % of queries. – Hallucination rate is demonstrably below 3 % on a held-out test set we w...
...relevant article link on our site (or an authoritative external document source if nothing in our library matches). We are open to whichever stack you prefer—OpenAI / ChatGPT, DeepAI, Perplexity, Anthropic, or a comparable LLM—provided you can prove hands-on experience building RAG workflows. Cost-effectiveness matters, so factor in token usage, caching, or lightweight vector stores such as FAISS, Pinecone, or similar to keep the running costs low. Key points you’ll handle • Connect the LLM to a vector index of our existing blog archive (about 1 000 posts) • Build or adapt a WordPress plugin/widget that captures the user prompt, calls the model, retrieves context, and returns the summary + link output in a friendly chat UI • Train or fine-...
...(PO, ASN, inventory, demand signals). Monitoring and observability: Experience instrumenting AI workloads for production reliability. Oracle Integration Cloud (OIC), Oracle API Gateway, or MuleSoft / Dell Boomi / Informatica for enterprise iPaaS. Hands-on experience with ServiceNow Flow Designer, IntegrationHub, or Virtual Agent for AI-enhanced ITSM workflows. vector databases (Pinecone, Weaviate, pgvector) and RAG architectures for enterprise document intelligence. AI governance, responsible AI practices, and enterprise LLM security (prompt injection, data leakage prevention). ...
...workflows using LangGraph, CrewAI, or similar frameworks • Develop RAG pipelines for university data retrieval • Implement persistent memory for long-term student tracking • Integrate backend with FlutterFlow or React frontend • Create semester-by-semester planning logic Required Skills: • Strong Python experience • LangChain / LangGraph • OpenAI or Claude API integration • Vector databases (Pinecone, Weaviate, Supabase Vector) • RAG architecture • AI agent orchestration Bonus: • Experience with MCP (Model Context Protocol) • Prior AI planner / assistant projects Project Details: • Remote work • Budget: $1000/month • MVP timeline: 4–6 weeks To Apply: Please send: • Portfolio...
...implementation that doesn't just post content, but acts as a backend manager handling tickets, payments, and deep user history across my website and social Tech StackAgent Framework: Hermes AgentSocial Orchestration: Postiz (Open Source)Integrations: Composio / Python / [Your CMS] API / Payment Gateways (e.g., Stripe)Control Interface: Telegram or Discord (Human-in-the-Loop)Database: Pinecone/SQLite (for Transactional & User History)Key DeliverablesDeployment: Full installation of Hermes on a dedicated server (VPS) with secure Automation: Configure the agent to manage customer support tickets, process emails, and track transactional Integration: Enable the agent to handle payments and refunds securely via API (e.g., Stripe/PayPal), with built-in
AI/ML & LLM Developer for Agentic AI System We are looking for an experienced AI/ML developer to build and deploy intelligent, agent-based systems using modern machine learning and large language model (LLM) technologies. Key Responsibilities. Design and develop AI/ML models...performance and scalability Handle data preprocessing, training, and evaluation Deploy solutions on cloud or local infrastructure Required Skills Strong experience in Python Hands-on experience with LLMs (e.g., OpenAI, open-source models) Experience with frameworks like LangChain / LlamaIndex / AutoGen Knowledge of machine learning & deep learning concepts Experience with vector databases (e.g., FAISS, Pinecone, Chroma) API integration and backend development skills Familiarity with cloud platforms (...
...scalable AI systems Design and optimize RAG pipelines Build and manage multi-agent workflows Improve latency, cost, and performance Implement evaluation frameworks for LLM outputs Work across backend, infrastructure, and AI layers Required Skills Strong Python + production experience Hands-on with LLMs (OpenAI, Anthropic, etc.) Experience with: RAG systems & embeddings Vector databases (FAISS, Pinecone, Weaviate, Milvus) LangChain / LangGraph / LlamaIndex Solid understanding of: APIs & microservices Distributed systems Nice to Have Multi-agent systems experience Fine-tuning / RLHF knowledge Real-time / async systems Evaluation metrics (NDCG, BLEU, etc.) Docker, Kubernetes, CI/CD Search / recommendation systems background What We’re Looking For Someone who thinks...
...transferred to a support rep with the full chat transcript. • All interactions are logged back into our SaaS dashboard for later analysis. Preferred stack is flexible—as long as the final solution plugs into the existing React front-end and can be maintained from our Node backend. Integrations with OpenAI or similar LLM APIs, LangChain for retrieval-augmented generation, and vector databases (Pinecone, Supabase, or comparable) are welcome suggestions. Acceptance criteria 1. Agent answers at least 80 % of common questions in staging tests with no hallucinations of non-existent features. 2. Handoff triggers correctly when confidence dips or when the user explicitly asks for human help. 3. Setup instructions and code are delivered in a clean repo with concise RE...
I need a production-ready chatbot that can sit behind the company firewall and answer product questions, search our knowledge base, and summarise internal documents. The core must be a Retrieval-Augmented Generation pipeline built with LangChain, LlamaIndex, or a comparable framework, and wired to a vector database such as Pinecone, Milvus, or FAISS for embedding storage. Python is the language of choice. I’m comfortable with either FastAPI or Flask, so feel free to work with whichever lets you move fastest. The LLM layer should default to GPT-4 Turbo for its accuracy, yet the codebase has to stay modular enough to swap in Qwen or ChatGLM when we need stronger Chinese support. Key pieces you’ll handle • Backend service that exposes clean REST endpoints for chat,...
...You should actively use: Claude / ChatGPT / Gemini Cursor / Copilot / AI coding tools Prompt engineering in real workflows Debugging and refining AI outputs Bonus (Very Relevant) RAG (Retrieval-Augmented Generation) Codebase indexing / embeddings Automated test generation Experience with logs / telemetry for feedback loops Tech Stack Python (preferred) Node.js / Java (good to have) Vector DBs (Pinecone, Weaviate, etc.) APIs, JSON, data pipelines What Stands Out (Real Signals) You’ve built something end-to-end using AI tools You rely on AI daily to write or improve code You’ve worked on real systems, not just notebooks... No companies to be apply and applicable only for individual freelancers...
...embeddings come in, relevant context is pulled in milliseconds, and the language model produces grounded answers. Here is what I need the finished solution to do: data ingestion, embedding creation, storage inside the chosen vector store, fast similarity search, and seamless hand-off of results to the LLM for final generation. I am open on stack (Python-based pipelines, LangChain, FAISS, Milvus, Pinecone, or another tool you are comfortable with), as long as it remains well-documented and reproducible on a standard cloud VM or container. Deliverables 1. Clean, modular source code for the ingestion–retrieval–generation pipeline 2. A configured vector database optimised for high-dimensional queries 3. REST or GraphQL endpoint (or CLI) that exposes a simple ask/a...
...medical literature, de-identified patient health records, and curated clinical-trial datasets. I already have secure access paths for each source; what I lack is the unified pipeline that cleans, embeds, and indexes the content so the model can ground every answer in verifiable evidence. Python, LangChain (or comparable orchestration), Hugging Face transformers, and a vector store such as FAISS or Pinecone feel like the natural toolkit here, but I am open to persuasive alternatives if they improve latency or compliance. HIPAA-level security and full audit trails are non-negotiable. Deliverables • Data-ingestion and cleansing pipeline connected to all three data sources • Vector index with citations back to the original documents or EHR entries • Fine-tuned ...
...can understand a question, pull the most relevant snippets from my curated algebra textbooks, formula sheets, and problem sets, then explain the answer step-by-step with rock-solid reasoning. Here is what success looks like from my side: • A working RAG stack (embeddings → vector database → LLM) wired together with modern tooling such as LangChain, LlamaIndex, or an approach you can justify. FAISS, Pinecone, Weaviate, or similar are fine for the vector layer; OpenAI, Claude, Llama 2/3, or comparable models may drive generation. • Retrieval tuned for high recall and low latency—students should see an accurate first response in well under a second for typical queries. • Step-wise explanations that a high-school learner can follow, with all reaso...
...(GHL)** - Upload and publish websites and landing pages within GHL - Build CRM pipelines, automations, and workflows for clients - Integrate third-party tools (including AI chatbots and voice tools) into GHL environments - Troubleshoot GHL issues across sub-accounts Additional desirable skills - Closebot, Retell, n8n, AI Architecture & Implementation, ElevenLabs - Experience with vector databases (Pinecone, Weaviate, or similar) for building RAG pipelines and AI knowledge bases for clients who need custom LLM solutions with grounded data retrieval. HTML / WEB DEVELOPMENT** - Upload websites and landing pages (HTML/CSS) to hosting environments - Make minor front-end edits: copy changes, layout tweaks, image swaps, button styling - Comfortable working with existing codebases...
... headers, body, query params. Support variable interpolation from conversation context or previous action outputs. - AI Generate — Call an LLM (OpenAI API) with a prompt template to generate/transform text. The prompt may reference variables. - Data Transform — Apply JSONPath expressions, map/filter data, restructure JSON payloads between actions. - Knowledge Lookup — Query a vector database (Pinecone) with a search query to retrieve relevant documents. - Send Email — Send an email via SMTP or API (e.g., SendGrid) with template variables. - Webhook / Notify — Send a webhook POST to a configurable URL with a JSON payload. - Human Escalation — Flag the conversation for human handoff, pause skill execution. - UI Prompt — Present a message or...
I need an experienced AI/ML consultant to architect and build a Retrieval-Augmented Generation (RAG) chatbot that streamlines internal workflows. The main goal is task automation: the bot should pull information from our proprietary data sources, reason over it with a large language model, and execute... Acceptance criteria – Bot reliably completes the agreed task scenarios end-to-end in our staging environment. – Latency under two seconds for typical queries. – No hallucinations on our benchmark set; fallback escalation when confidence drops. – All code, configs, and instructions compile and run on a fresh machine. If you have recent hands-on experience with RAG, vector DBs such as Pinecone/FAISS, and enterprise integrations, I’m ready to revi...
... CrewAI, AutoGen, or similar) Prompt engineering skills — writing effective system prompts, managing context windows, and building knowledge bases for LLMs Experience with HubSpot API (deals, contacts, tickets, pipelines, workflows) and API Experience building agents or bots that automate multi-step business processes (not just chatbots) Familiarity with vector databases (ChromaDB, Pinecone) and RAG architectures HTML/CSS for building dashboards and reports Experience working with a small team remotely across time zones Nice-to-Have: Python scripting (some of our automation uses Python) Experience with Pipedream or serverless deployment Familiarity with Anthropic's Claude API or OpenAI API Background in business process automation (not just coding — understandi...
...frameworks such as LangChain, LlamaIndex, or similar. - Integrate LLMs (e.g., GPT-4, Claude) for reasoning, generation, and tool usage. - Build multi-step reasoning pipelines (planning, memory, retrieval, execution). - Develop and manage RAG (Retrieval-Augmented Generation) systems. - Connect agents with external tools (APIs, databases, search engines). - Implement vector search systems (FAISS, Pinecone, ElasticSearch). - Design short-term and long-term memory systems for agents. Backend & System Development: - Develop scalable backend services using Python (FastAPI or Flask). - Build APIs to support agent interactions and orchestration. - Optimize system performance, latency, and cost of LLM usage. - Ensure robust error handling and fallback strategies for AI workflows. F...
...pattern. The core service runs on Spring Boot, pulls context from a Pinecone vector index, streams prompts to OpenAI, and returns concise text-based answers. All application data and user sessions live in a MySQL database. What must be in place • Secure user authentication so each visitor has a private history. • Multilingual support—English to start, with the structure in place to add more languages through locale files and prompt templates. • Basic analytics and reporting: conversation counts, average latency, most-requested topics, and simple charts exposed via a protected REST endpoint or lightweight dashboard. Flow I expect 1. A user signs in. 2. Their query is vectorised, relevant chunks are retrieved from Pinecone, and combined with ...
...scalability of the solution * Demonstrate real-world use cases such as Q&A systems, document search, or intelligent assistants **Scope of Work:** * Understand project requirements and define system architecture * Design and implement a RAG pipeline (retriever + generator) * Prepare and preprocess datasets / documents (PDFs, text, structured data, etc.) * Implement vector database (e.g., FAISS, Pinecone, Chroma, etc.) * Integrate embedding models and LLM (OpenAI, open-source models, etc.) * Develop and train/tune models where necessary * Evaluate system performance (accuracy, relevance, latency) * Build a simple demo interface or API to showcase functionality * Document the entire workflow and architecture **Deliverables:** * Working RAG-based AI/ML prototype * Source code...
...plugin, admin panel, dark mode, WCAG 2.1 AA accessibility. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ TECH STACK • Frontend: React/ + TypeScript, Tailwind, WebSocket • Backend: Node.js or Python (FastAPI), event-driven async, message queue • AI/LLM: LangChain, Claude API (primary), GPT-4 (fallback), third-party detection APIs • Database: PostgreSQL, Redis, Elasticsearch, Vector DB (Pinecone/Weaviate) • Infra: Docker/Kubernetes, AWS or GCP, Terraform, GitHub Actions CI/CD • Security: OWASP Top 10, TLS 1.3, AES-256, zero-retention processing, pen test before launch • Performance: Page load ≤2s, detection ≤60s video / 30s audio / 10s text, 99.5% uptime ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60-DAY SPRINT TIMELINE Phase 1 &md...
