Lucky Anish

AI/ML Engineer
Agentic Systems
2026
PORT
FO
LIO
Lucky Anish

Bio

Final-year B.Tech (AI/ML) student at NIMS University who ships measurable, production-oriented systems — spanning computer vision, NLP/RAG, and agentic LLM tool-calling. National Finalist in QuizOff 2026.

Education

B.Tech — AI & ML
NIMS University, Jaipur
Artificial Intelligence & Machine Learning
Coursework: Deep Learning, NLP, CV, DSA
(2023 - Present) • Expect to graduate in 2027
Class XII (Higher Secondary)
Gurukripa Academy, Kahalgaon
Subjects: Physics, Chemistry & Mathematics
Aggregate Score: 84%
Class X (Secondary)
St. Xavier's School, Sahibganj

Expertise

Python
PyTorch / TensorFlow
LangChain & RAG
OpenCV & YOLOv8
FastAPI & Node.js
Docker & Git

Languages

English • Professional
Hindi • Native

Certifications

JPMorgan Chase & Co. / Forage
Software Engineering Simulation (Jun 2026)
Code: 6a27e3a15b0bc151b65cfa6a
Anthropic MCP Certified
ID: 5utqn2pfrk7n
NVIDIA Deep Learning Inst.
ID: jjzDXD9kS3u4IOXSbhuMUw
Google Cloud GenAI
ID: 9638839

Work Experience

InAmigos Foundation (IAF) (Jun 2026 - Jul 2026)
Jaipur, Rajasthan, India • Remote
AI Web Development Intern
  • Engineered AI-driven web development modules, intelligent user interfaces, and API endpoints.
CodeAlpha (Sep 2025 - Oct 2025)
Virtual Internship Program
Machine Learning Intern
  • Built a disease prediction model using patient medical data and classification algorithms (Scikit-Learn).
  • Built a credit score prediction model applying supervised learning on financial data.

Featured Projects & Open Source

Agentic AutoML Framework (GitHub ↗) (Aug 2026 - Present)
Advisor Design Pattern System
  • Architected explainable AutoML framework logging structured rationale for every pipeline stage with full CI.
Identity Reconciliation Service (GitHub ↗) (Apr 2026)
Backend Service
  • Reconciles customer contact records across orders (13/13 Jest tests passing).
Kalman Filter Multi-Object Tracker (GitHub ↗) (Jun 2026 - Present)
First Principles NumPy Implementation
  • SORT-style tracker cutting RMSE by 31% and jitter by 86% over Euro Filter baseline.
Full-Stack AI Room Visualizer
  • Built & deployed web application featuring a real-time AI-powered room visualizer for interior room including hall, rooms, Kitchen, Bathroom.

Leadership & Achievements

QuizOff 2026 — India's Biggest AI Quiz (Jul 2026)
National Finalist
  • Competed among 5,25,000+ students from 48,500+ institutions across 35+ countries.
Agentic-IQ GitHub Organization (Apr 2026 - Present)
Open-Source Contributor
  • Active open-source contributor building agentic tool-calling modules.
AutoML Pipeline Thumbnail
Key Highlights
  • a Automated Data Preprocessing
  • b Intelligent Feature Engineering
  • c Algorithm Selection & Tuning
  • d Model Comparison & Deployment
01

AGENTIC
AUTOML
FRAMEWORK

End-to-End Automated Machine Learning Pipeline / 2026
View on GitHub ↗
Agentic AutoML Architecture
System Workflow
1 Data Analysis, Cleaning & Feature Engineering
2 Model Selection, Tuning, Comparison & Deployment

The Agentic AutoML Framework takes a raw dataset and automatically executes the entire machine-learning pipeline with minimal manual intervention. The intelligent workflow seamlessly handles Dataset Analysis, Cleaning, Feature Engineering, and Algorithm Selection. It trains and evaluates multiple candidate models, automatically identifies the best performer, and saves the complete pipeline for immediate deployment and prediction.

Tools Used: Python, PyTorch, LangChain, Scikit-Learn, Docker, GitHub Actions CI
02

IDENTITY
RECON
CILIATION
SERVICE

Backend Customer Data Linker / 2026
View on GitHub ↗

The Identity Reconciliation Service is an automated backend API that identifies and consolidates customer identities across varying email and phone combinations. Built with Node.js, Express, and Prisma ORM, it exposes a POST /identify endpoint that intelligently processes incoming contacts. It maintains primary and secondary relationships, and safely merges conflicting identities using database transactions to ensure data consistency.

Tools Used: Node.js, Express, Prisma ORM, SQLite, Jest
Identity Reconciliation Architecture

Built using Node.js, Express, Prisma ORM, and SQLite, the service exposes a POST /identify API that analyzes incoming contact information and handles new identities, exact matches, partial matches, and conflicts between previously separate customer groups.

The system maintains primary and secondary contacts, automatically merges conflicting identities by keeping the older primary record, and re-links associated secondary contacts. Database transactions are used during merges to ensure data consistency.

Key Highlights:
  • Automated customer identity reconciliation
  • Primary/secondary contact relationship management
  • Handling of new, exact, partial, and merge scenarios
  • Transaction-safe database operations
  • REST API using Express & Prisma
  • Jest unit tests covering core business cases
Kalman Filter Tracking Up
Key Highlights
  • a YOLOv8 Object Detection
  • b Kalman Filter State Prediction
  • c IoU & Hungarian Matching
  • d Track Management & Updates
03

KAL
MAN
FILTER

Real-Time Object Tracking System / 2026
View on GitHub ↗
Kalman Filter Tracking Down
System Workflow
1 Object Detection (YOLOv8)
2 State Prediction (Kalman Filter)
3 Matching (IoU & Hungarian Algorithm)
4 Track Updates & Maintenance

The Kalman Filter Multi-Object Tracker is a real-time system that detects and maintains track of multiple objects across video frames. Even when detection is noisy or temporarily missing, the system predicts where each object will move using a mathematical state estimation filter.

The tracker automatically creates new tracks when objects appear and continuously predicts positions during occlusions. It is benchmarked against the One Euro Filter, demonstrating superior handling of jitter and missed detections.

Tools Used: Python, NumPy, YOLOv8, OpenCV, Matplotlib
UmaTraderss

UmaTraderss (umatraderss.com) is a luxury e-commerce & web application built for premium Makrana marble and tiles. To connect customers seamlessly to luxury products, a real-time AI-powered room visualizer is integrated into the web application. The system also features multi-room selection enabling customers to preview flooring across living halls, bedrooms, kitchens, and bathrooms.

Tools Used: React, Next.js, Node.js, FastAPI, Real-Time AI Engine, Tailwind CSS, Vercel
04

UM
ATRA
DERSS

Full-Stack AI Room Visualizer & E-Commerce / 2026
Visit umatraderss.com ↗
Key Highlights
  • a Real-Time AI Flooring Visualizer
  • b Multi-Room Selector (Hall, Kitchen, Bath)
  • c Luxury Makrana Marble Catalog
  • d Instant Inquiry Engine
Thank you.