Headline

Artificial Intelligence Engineer @ Samsung | Building AI Solutions

About

I am a performance-driven AI Engineer with 3 years of experience, currently building LLM-powered systems at Samsung Electro-Mechanics. My work focuses on turning applied AI research into reliable, production-grade tools that engineering teams use daily. My technical focus lies in designing retrieval-augmented generation (RAG) pipelines, agentic workflows, and fine-tuned language models that solve real engineering problems — from semiconductor datasheet search to automated ticket triage — while keeping every system auditable, scalable, and safe for production use. Key Highlights: Recognized Excellence: Awarded "Best Employee" (H1 2024) at Problem Solving: Secured 1st place at CentuRITon, a national-level hackathon, competing against 100+ teams and endorsed by Devfolio & MLH. Measurable Impact: Cut average issue resolution time nearly in half (40 min → ~22 min) with LangGraph-based automation agents, and improved retrieval precision by ~20% through custom metadata-aware document chunking. Core Technologies: Python, LangChain, LangGraph, HuggingFace Transformers/PEFT, RAG, Elasticsearch, FastAPI, PostgreSQL, Docker, Kubernetes, MLflow, gRPC

Location

Greater Bengaluru Area

AI stack

claude

codex

cursor

gemini

hugging face

ollama

Work experience

Artificial Intelligence Engineer at Samsung Electro-Mechanics, Aug 2023 - Present · 3 yrs 2 mos, Bengaluru, Build and maintain internal RAG pipelines using LangChain, Elasticsearch (hybrid dense + BM25 search), and OpenAI embeddings, enabling engineers to query semiconductor datasheets and technical documents. Design document ingestion pipelines with metadata-aware chunking to improve retrieval precision and search relevance across large internal document corpora. Develop LLM-based agents using LangGraph with ReAct prompting to automate internal workflows such as log analysis and ticket triage, integrating tools for SQL querying, REST APIs, and file parsing. Implement human-in-the-loop guardrails in agent workflows, requiring engineer approval before destructive actions, with full execution traces logged for auditability. Fine-tune open-source LLMs (Mistral-7B) using QLoRA on domain-specific data, running inference entirely on-prem. Set up LLM evaluation workflows using RAGAS, integrated into CI pipelines to catch prompt and model regressions before deployment. Deploy model inference services on an on-premise Kubernetes cluster using Docker, Helm, and Nginx, with auto-scaling to handle production traffic. Own the MLOps lifecycle for internal ML projects — dataset/model versioning, automated training pipelines, and staged deployment rollouts. Skills used: Python · LangChain · LangGraph · Elasticsearch · HuggingFace Transformers · PEFT/QLoRA · RAG · FastAPI · Docker · Kubernetes · MLflow · PostgreSQL · CI/CD

Software Engineer at Samsung Electro-Mechanics, Feb 2023 - Aug 2023 · 7 mos, Bengaluru, Karnataka, India

Intern at The Machine Learning Company, Nov 2021 - Jan 2022 · 3 mos, India, During this period I completed three projects under the guidance of the TMLC. The projects are: 1: Agricultural N2O Flux prediction 2: Patient survival prediction 3: Reddit upvote prediction using machine learning

Fellowship at Alphaa AI, Dec 2020 - Feb 2021 · 3 mos, Newark, Delaware, United States

Education

Bachelor's degree, Computer Science, Dayananda Sagar College of Engineering, BANGALORE

12, Mathematics, kendriya vidyalaya

Public links

github: https://github.com/navahchseham

navahchseham-portfolio.netlify.app: https://navahchseham-portfolio.netlify.app/