Likith Shetty

M.Sc. Artificial Intelligence & Machine Learning, batch 2025 to 2027 CHRIST (Deemed to be University), Bengaluru

likithshetty188@gmail.com · github.com/LikithSh3tty · linkedin.com/in/likith-shetty

Summary

AI and machine learning postgraduate and full-stack developer. I build and ship complete products on my own, mostly around agents, retrieval and applied ML. Looking for a software or AI engineering role focused on building practical products rather than demos.

Experience

Software Development Intern, PayNearby, Mumbai Jun 2024 to Jul 2024

• Developed and tested applications in C and Python, debugging code-level issues to improve reliability.
• Built frontend interfaces in HTML, CSS and JavaScript, and integrated and tested APIs to fetch and display data.
• Used Git for version control and SQL for database operations.

Projects

Agenvo · React, Firebase, LangGraph, FastAPI agenvox.vercel.app

• Multi-tenant agency income tracker with a guided onboarding wizard, industry presets and config-driven terminology.
• Commission engine covering percentage, flat, tiered and hourly rules, with per-client configuration and automated agency and team split calculation.
• Built-in assistant on a LangGraph state graph. Every figure is computed deterministically in Python and the model only phrases facts it is handed, so it cannot invent a number.

CloudNest · Python, LangGraph, FastAPI, React cloudnest-nine.vercel.app

• Support agent that answers from product documentation by meaning rather than keywords, using a local embedding model that runs inside the serverless function at zero per-query cost.
• Parallel semantic and lexical retrieval fused with Reciprocal Rank Fusion, behind a calibrated confidence gate.
• Escalates to a Postgres ticket queue with a read-only admin view when it is not confident or the user asks for a human.

DriftBell · LangGraph, n8n, FastAPI, SQLite

• ML drift watchman that investigates a model before it retrains it and waits for human approval.
• Cyclic agent with tool use and bounded self-critique, checkpointed to SQLite so a graph parked on a decision survives a restart.
• Pairs n8n’s Wait node with LangGraph’s interrupt(), so an approval arriving hours later still lands on a live graph.

Grid0pt · Python, FastAPI, Shapely, OpenCV, React

• Grid packing optimizer that sweeps every fractional offset and rotation to fit the most whole cells into an irregular polygon.
• Extracts boundaries and obstacles from uploaded floor plans or hand-drawn sketches via contour-hierarchy analysis.

Writing

Why Small Language Models Might Be the Future Medium, March 2026

On the case for models small enough to run locally, and why the interesting frontier may be downward rather than upward.

Skills

Languages Python, JavaScript, C Web React, Node.js, Astro, Svelte, HTML, CSS AI and ML PyTorch, scikit-learn, TensorFlow, LangGraph, Pandas, NumPy Data Firebase, Supabase, Postgres, SQL Tools Git, GitHub, Vercel, VS Code, Figma

Education

CHRIST (Deemed to be University), Bengaluru M.Sc. Artificial Intelligence & Machine Learning, expected 2027

Nitte University, Mangalore Bachelor of Computer Applications, 2022 to 2025. CGPA 9.61 / 10.0

Leadership

Club Leader, IT Club, Nitte University, Mangalore 2023 to 2024

Led the university IT Club, organising events, hackathons and workshops for the IT department.

Certifications

• Programming with JavaScript, Meta / Coursera
• HTML and CSS in Depth, Meta / Coursera
• Version Control, Meta / Coursera
• Introduction to Front-End Development, Meta / Coursera
• Python Core, Sololearn