Skip to content
View Divyansh-git10's full-sized avatar

Block or report Divyansh-git10

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Divyansh-git10/README.md

Divyansh Gautam

AI Engineer · GenAI · LLMs · RAG · Agentic Systems · Computer Vision


🚀 About

I'm an AI Engineer who builds production-oriented AI systems — LLM, RAG, and multi-agent pipelines — with a solid computer-vision background. I care about the engineering behind the models: systems that are evidence-grounded, tested, and genuinely deployable — not just notebooks.

My professional experience spans production LLM and multi-agent systems, voice-to-voice assistants, and document-intelligence pipelines. Much of that work is client- or employer-owned and isn't public — the repositories here apply the same engineering standards to projects I can share openly.


🛠️ Technologies I work with

  • GenAI & LLMs — OpenAI API · Ollama · OpenRouter · Groq · Gemini · RAG · DSPy · prompt engineering
  • Agentic Systems — multi-agent orchestration · tool use · CrewAI · n8n · evaluation & validation
  • Retrieval & Vector Search — FAISS · Pinecone · Sentence-Transformers
  • Document AI — Azure Document Intelligence
  • Computer Vision — YOLOv11 · ByteTrack / BoT-SORT · OpenCV · TensorFlow / Keras
  • Backend & MLOps — FastAPI · SQLAlchemy · Streamlit · Docker · pytest · GitHub Actions · Railway
  • Languages — Python · JavaScript

⭐ Selected Work

A couple of highlights from my public repositories — the full set is pinned below:

  • FitNova — Sales-Call Intelligence — a local-first speech→LLM analytics platform (transcription, diarization, evidence-grounded scoring and issue detection) exposed through a FastAPI API, a Typer CLI, and a Streamlit dashboard over one shared data layer, backed by a full test suite.
  • Suproc AI Agent — a safe, deterministic agentic workflow whose validator independently re-derives every recommendation from the source data, with prompt-injection handling and a mandatory human-approval gate.

↳ More — a deployed RAG + memory agent, a production RAG service, a multi-agent document-QA system, and computer-vision work — in the pinned repositories below.


📬 Connect

Pinned Loading

  1. FitNova-AI-Sales-Intelligence FitNova-AI-Sales-Intelligence Public

    Local-first sales-call intelligence: transcription, diarization, evidence-grounded LLM scoring & issue detection, exposed through a FastAPI API + Typer CLI + Streamlit dashboard over one repository…

    Python

  2. Suproc-AI-Agent Suproc-AI-Agent Public

    Local, deterministic agentic workflow that turns a natural-language sourcing request into validated supplier recommendations — an independent validator re-derives every fact from the database, with…

    Python

  3. vestaff-rag vestaff-rag Public

    Retrieval-Augmented Generation over the AWS Customer Agreement: PDF → recursive chunking → MiniLM embeddings → FAISS retrieval → Groq LLM with a strict grounding prompt, plus SQLite query logging a…

    Python

  4. Soccer-Player-Reid-YOLOV11-Bytetrack Soccer-Player-Reid-YOLOV11-Bytetrack Public

    Real-time soccer player tracking & re-identification with YOLOv11, plus a benchmarked comparison of ByteTrack, BoT-SORT, DeepSORT, and Norfair on 720p/4K footage. Python · OpenCV · Ultralytics.

    Jupyter Notebook 1

  5. persistent-sales-agent persistent-sales-agent Public

    Deployed FastAPI sales assistant built as an agentic workflow: tool-orchestrated semantic catalog search (MiniLM + FAISS), persistent cross-session memory, and LLM self-evaluation. Live on Railway.

    Python

  6. Agentic-RAG-Chatbot Agentic-RAG-Chatbot Public

    Multi-agent RAG chatbot: Ingestion → Retrieval → LLM-Response agents exchanging structured messages, with multi-format parsing (PDF/DOCX/PPTX/CSV/text), MiniLM+FAISS retrieval, and a Streamlit UI.

    Jupyter Notebook 1