Yash Sharma.
Building production ML systems that operate in the real world, not just Jupyter notebooks. NLP, GenAI, and full-stack deployment.
The Engineer
Behind the Model.
I'm an early-career AI/ML Engineer who builds systems that actually run in production. Not prototypes that impress in demos — real services with latency SLAs, test coverage, CI/CD pipelines, and monitoring dashboards.
At Celebal Technologies I owned the entire NLP pipeline end-to-end — raw data through feature engineering, model training and evaluation, FastAPI deployment, CI/CD automation, and Streamlit observability. That's the standard I hold every project to.
Currently pursuing my MCA while actively building AI-native applications. I specialize at the intersection of rigorous ML engineering and production software craftsmanship — the gap most data scientists leave unfilled.
Professional Track Record.
Data Science Intern
- Built LDA topic modelling across 50,000+ records paired with a Streamlit observability dashboard — collectively reducing manual triage effort by 40%.
- Delivered ownership across the full ML stack — raw data, features, model evaluation, REST API, CI pipeline, and monitoring.
Teammate — Operations
- Maintained 95%+ CSAT consistently, ranking among top performers.
- Reduced repeat cases by 15% through data-driven issue pattern analysis.
Customer Service Representative
- Conducted root-cause analysis of high-frequency issue patterns, reducing repeat contacts by 15% and improving first-contact resolution.
Selected Work.
Customer Inquiry Classifier
Production NLP pipeline classifying support tickets into 7 categories. Real-time inference served via FastAPI. Full CI/CD, automated testing, and Streamlit observability dashboard. Built to enterprise standards.
AI Trip Planner
GenAI itinerary platform built during Google Gen AI Exchange Hackathon 2025, selected for Phase 1. Multi-step prompt orchestration, real-time data fusion, and streaming for a low-latency interactive planning experience.
DocuMind AI Copilot
AI-powered document copilot — upload documents and chat with them via an LLM-powered RAG pipeline. Supports multi-doc querying with context-aware responses.
Adaptive Diagnostic Engine
LLM-powered IT support automation agent with adaptive questioning, context memory, and multi-step resolution flows. Browser-based agentic architecture for real-world diagnostic workflows.
Technical Arsenal.
Core ML
Engineering
LLM & GenAI
Awards & Recognition.
Google Gen AI Exchange Hackathon 2025
Built production GenAI Trip Planner with Gemini on Vertex AI. Demonstrated real-time LLM streaming, prompt orchestration, and full-stack AI architecture.
Prime AI/ML Batch — Apna College
Intensive program Python, ML, Deep Learning, GenAI, AI Engineering Stack. Practical, project-driven curriculum delivering real-world model deployment skills.
GenAI Mastermind — Outskill
Advanced LLM workflows, AI agents, and automation. Prompt engineering, agentic architectures, and production LLM system design.
Peer Recognition.
Yash delivered more than expected. The NLP pipeline was production-ready on day one — clean code, full test coverage, and latency that surprised the entire team. He thinks like an engineer, not just a data scientist.
What impressed me was his ownership. He didn't wait for instructions — he identified the observability gap, built the dashboard, and improved the deployment pipeline. That's someone who goes well beyond the brief.
Academic Foundation.
Master of Computer Application
Bachelor of Computer Application
Certifications & Programs
Let's Build
Something
Excellent.
I'm actively seeking AI/ML engineering roles, internships, and contract work — particularly in NLP, production ML systems, and LLM-powered applications. Open to remote positions globally.
If you're building something worth building, I'd like to talk.
Send a Message →