Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Appearance settings
View MansiDakhale's full-sized avatar
  • IIITL

Block or report MansiDakhale

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
MansiDakhale/README.md

Hi there, I'm Mansi 👋

AI Engineer | MSc AIML @ IIIT Lucknow | Currently @ Evomaton

I bridge the gap between complex machine learning research and scalable production systems. I specialize in building agentic workflows, rigorously evaluating RAG pipelines, and engineering the backend infrastructure (FastAPI, Celery, AWS) to serve AI efficiently.

Featured Work

  • Architecture: Engineered a multi-agent platform using LangGraph and Groq (LLaMA 3.3 70B) with a Planner agent routing queries for RAG and tool execution.
  • Backend & Infra: Built robust async ingestion via Celery + Redis and real-time token streaming (SSE) to a React frontend. Deployed end-to-end on AWS EC2/RDS and containerized with Docker.
  • Research: Designed a novel Uncertainty-Guided Soft-Attention Block (UG-SAB) using Monte Carlo Dropout to synthesize dynamic contrast-enhanced MRIs.
  • Impact: Cut annotation needs by 40% via pseudo-labelling and achieved +6.5% Global SSIM over fully-supervised baselines.
  • Data: Meticulously curated and preprocessed a paired dataset from 239 patients in the I-SPY1 dataset.
  • Evaluation & Metrics: Implemented an LLM-as-a-judge (RAGAS) evaluation pipeline scoring faithfulness and context precision. Achieved 92.4% retrieval precision@5 and 84.7% keyword accuracy.
  • Tech Stack: LangChain, ChromaDB, HuggingFace embeddings, and fully offline inference via Ollama/LLaMA.
  • Workflow: Automated end-to-end ecommerce creative generation using a 6-stage multi-agent LangGraph workflow.
  • Engineering: Built FastAPI APIs with Celery + Redis for async job execution and CSV-based bulk processing.

Core Tech Stack

  • AI Systems: LangGraph, RAG, Prompt Engineering, LoRA/QLoRA, Agentic AI, Groq
  • Deep Learning: PyTorch, GANs, ViT, CNNs, Hugging Face
  • Backend & MLOps: FastAPI, Docker, Celery, Redis, AWS (EC2/RDS), W&B
  • Languages: Python, SQL

Let's Connect

Popular repositories Loading

  1. New_Movie_Recommendation_System New_Movie_Recommendation_System Public template

    Python

  2. Fake_News_Detection Fake_News_Detection Public

    Python 1

  3. generative-ai-for-beginners generative-ai-for-beginners Public

    Forked from microsoft/generative-ai-for-beginners

    21 Lessons, Get Started Building with Generative AI 🔗 https://microsoft.github.io/generative-ai-for-beginners/

    Jupyter Notebook

  4. Agentic-RAG-Powered-AI-Research-Assistant Agentic-RAG-Powered-AI-Research-Assistant Public

    Agentic RAG pipeline using LangChain, ChromaDB, and local HuggingFace/Ollama models for intelligent document research.

    Python

  5. frs_project frs_project Public

    Python

  6. TSGAN-Project TSGAN-Project Public

    Python

Morty Proxy This is a proxified and sanitized view of the page, visit original site.