Hi, I am Manas Tiwari 👋

I am a Software Engineer with 2+ years at Siemens Healthineers building AI/LLM systems, automation frameworks, and CI/CD pipelines. Specialized in C++, Python, and C# with hands-on experience in RAG pipelines, LLM fine-tuning (Mistral 7B), and distributed systems. M.S. Computer Science — Machine Learning, University of Tennessee.

Experience

Siemens Healthineers

Software Engineer / Automation  ·  C#, .NET, Python, Jenkins, LangChain | Jan 2024 – Present
  • Built end-to-end RAG pipeline using LangChain, HuggingFace Transformers & ChromaDB; fine-tuned Mistral 7B with LoRA/QLoRA for medical domain tasks.
  • Replaced legacy UI automation with direct API calls — 75% reduction in CI test execution time, saving 50+ engineering hours weekly.
  • Designed Jenkins CI/CD across 10 distributed Linux/macOS nodes, achieving 30% reduction in build time through parallelization.
  • Built automated analyzer for 750+ unit/integration tests, improving code quality by 30%.

University of Tennessee

Graduate Teaching Assistant  ·  C++, Python | Aug 2022 – Dec 2023
  • Built automated grading system in Python/Shell — 40% faster grading, 20% accuracy improvement.
  • Implemented distributed parallel computing with MPI/OpenMPI across multi-node environments, improving compute performance by 30%.

Cognizant

Programmer Analyst Intern  ·  C++, Azure DevOps | Apr 2022 – Jul 2022
  • Reduced deployment time by 20% through CI/CD automation with Jenkins and Azure DevOps in an Agile environment.

My Projects

AI / LLM  ·  2025
Domain-Specific RAG Chatbot

  • Designed distributed architecture separating vector storage (ChromaDB), embedding generation, and LLM inference into independently deployable services.
  • Fine-tuned Mistral 7B with LoRA/QLoRA — reduced hallucination rates and improved domain accuracy over the base model.
  • Implemented retrieval chains, prompt templates, and evaluation loops using LangChain and e5-large-v2 embeddings.
Machine Learning  ·  2023
Sign Language Recognition — CNN
CNN screenshot

  • Real-time gesture recognition pipeline in C++ / OpenCV for detection and feature extraction.
  • Trained CNN in Python / TensorFlow achieving 90%+ accuracy. Awarded 3rd prize — Best Project.
Full Stack  ·  ReactJS
movies.io
movies.io screenshot

Movie browsing web app using ReactJS and the TMDB API — fetches real-time data on latest and upcoming movies with ratings and reviews.