NK
Initializing Portfolio000%
Portfolio — 2026

NutankumarKM

AI Engineer✦Systems Builder

I build production AI systems — RAG pipelines, autonomous agents, and cross-platform products that ship.

Say Hello
120ms
RAG query latency
4
AI Agents Shipped
8.0
B.E. CGPA / 10
150ms
Raft commit latency
02Who

About

nutan@portfolio:~/aboutonline
KM Nutankumar
profile.jpg1024×1280 · rgb
1
v2026.1

I'm a Computer Science & Design graduate from Tontadarya College of Engineering, currently building AI-based learning management systems at KAM Global AI. My work lives at the seam between research-grade ML and shippable software — production RAG pipelines, fine-tuned open-source LLMs, autonomous LangChain agents, and cross-platform Flutter applications.

What I care about is end-to-end ownership. A 120ms retrieval latency is only useful if it survives production traffic. A fine-tuned LLaMA 3 is only useful if the inference pipeline stays under budget. I obsess over both halves: the model that reasons, and the system that serves it.

Beyond AI, I'm drawn to distributed systems fundamentals — I once built a 5-node Raft key-value store from scratch with leader leases, log snapshotting, and crash recovery, just to understand consensus at the level of the wire protocol. That curiosity drives everything I ship.

Education

B.E. in Computer Science and Design

Tontadarya College of Engineering, Gadag

2022 — 2026
8.0 / 10.0
Download Resume↓
03Toolkit

Skills

Python✦LangChain✦LLaMA 3✦Mistral✦RAG✦Flutter✦HuggingFace✦PEFT✦LoRA✦MCP✦Vector DB✦OpenCV✦Docker✦ZeroMQ✦Raft✦Firebase✦MongoDB✦TypeScript✦Python✦LangChain✦LLaMA 3✦Mistral✦RAG✦Flutter✦HuggingFace✦PEFT✦LoRA✦MCP✦Vector DB✦OpenCV✦Docker✦ZeroMQ✦Raft✦Firebase✦MongoDB✦TypeScript✦
AI / MLLLM & RAGApplication DevLanguagesData & CloudTools & Craft
01
TensorFlow
02
Scikit-learn
03
NLP
04
Pandas
05
NumPy
06
Generative AI
07
Prompt Engineering
08
AI Integration
6
Categories
40+
Tools & libs
7
Languages
4
Domains
03.5 / Suitability

Job Fit Analyzer

Recruiter shortcut: paste a job description below to instantly check compatibility, matched skill sets, and relevant projects.

Paste Job Description

Analyzed locally in your browser — nothing is uploaded.

Or load a preset job description:

Awaiting Job Description

Paste text or click one of the quick presets on the left to analyze candidate compatibility dynamically.

04Trajectory

Experience

Jan 2026 — PresentRemote

AI Engineer

KAM Global AI

AI-based LMS division — intelligent learning systems and cross-platform AI product development.

  • Developed and deployed a production RAG pipeline achieving 120ms average query latency using vector databases and domain-specific knowledge bases.
  • Built 4 autonomous AI agents using LangChain with memory chains, prompt templates, and tool integrations for production LMS features.
  • Developed cross-platform mobile and web applications with integrated AI capabilities using Flutter and REST APIs.
  • Fine-tuned LLaMA 3 and Mistral using PEFT and LoRA for task-specific performance in LMS workflows.
  • Designed AI-powered LMS modules including intelligent Q&A, content summarization, and adaptive learning flows.
LangChainLLaMA 3MistralPEFTLoRAFlutterRAGVector DB
05Selected

Projects

202501 / 04

MCP-Powered AI Developer Assistant

A local AI dev assistant that reads codebases and ships actions.

Built using Anthropic's Model Context Protocol, the assistant connects to GitHub, the file system, and databases as MCP servers for unified tool access. A RAG pipeline lets it reason over entire codebases, create issues, query databases, and trigger autonomous actions.

3+
MCP servers
Multi-step
Agent type
PythonMCPLangChainRAGGitHub APIClaude API
202502 / 04

Fine-Tuned Domain LLMs

LLaMA 3 + Mistral fine-tuned for medical & legal domains.

Parameter-efficient LoRA training on custom instruction datasets for medical and legal domains. Built high-quality instruction data, evaluated on domain benchmarks, and shipped scalable RAG-augmented inference at reduced compute cost.

PEFT / LoRA
Method
Medical, Legal
Domain
PythonHuggingFacePEFTLoRALangChainLLaMA 3Mistral
202403 / 04

AI Proctoring System

Real-time live proctoring with gaze tracking & malpractice flags.

Real-time AI proctoring system with live camera face detection, gaze tracking, and candidate presence verification. Multiple-face detection, browser tab-switch alerts, and suspicious-behaviour flagging via OpenCV and MediaPipe. Flutter frontend, Firebase backend for session logging and reporting.

Face + Gaze
Tracking
Flutter
Frontend
PythonOpenCVMediaPipeComputer VisionFlutterFirebase
202404 / 04

Raft Distributed Key-Value Store

5-node Raft cluster with leader leases, snapshots, and failover.

5-node Raft-based distributed key-value store with leader leases, log snapshotting, and crash recovery. Achieved under 150ms commit latency and 100% recovery success rate with automatic failover and real-time monitoring.

<150ms
Commit latency
100%
Recovery
PythonZeroMQDockerRaft Consensus
Honors
5G / 6G

IIITB COMET Foundation Internship

Selected competitively for the FWC Project Internship on 5G and 6G wireless systems research — stipend awarded.

Healthcare ML

National Hackathon — Healthcare ML

Designed and delivered healthcare and insurance ML prediction models at a national hackathon, achieving 70–80% accuracy.

Collaboration

Let's build something rare.

Direct Contact

I am open to full-time AI engineering roles, high-complexity contract builds, or research collaborations in distributed consensus.

Consulting Planner

Scope & Estimate Project

Step 1 of 3
  • Type
  • Scale
  • Target
  • Scope
What are we building?