AI & COMPUTER VISION DEVELOPER

|
Kolkata, India· B.Tech CSE '27· ex-Samsung R&D
Currently building — IGNIS Phase 2

"Every line of code begins with a question worth answering."

sourav_burman.py
class SouravBurman:

  def __init__(self):
    self.role  = "AI Developer"
    self.uni   = "Brainware University"
    self.batch = "2023 – 2027"
    self.sgpa  = 8.79
    self.past  = "Samsung R&D, Kolkata"
    self.goal  = "Build what matters"

  def obsession(self):
    return "Real problems → impact"

I'm a third-year Computer Science student at Brainware University, Kolkata (SGPA 8.79) — specialising in AI and Computer Vision.

I've built systems that detect eye conditions without hardware, classify live traffic in real time, read market sentiment from raw text — and a complete personalised AI fitness platform powered by Claude AI.

I trained in Cryptographic Security at the Indian Statistical Institute, Kolkata under a MeitY programme — and completed Qualcomm's Deep Tech Entrepreneurship course with a perfect 100% score.

I'm not interested in average things. I'm obsessed with the intersection of precision engineering and real human impact.

5Live Projects
6Certifications
∞Problems to Solve

"Shipped is better than perfect. But I ship both."

Things I've built.

01

Ocular AI v2

Problem: Refractive error detection requires expensive equipment inaccessible to billions.

Software-only smartphone screening using photorefraction — n=92, r=0.9857, avg error 0.241D, 85.9% within ±0.5D.

Proprietary research — no hardware required. OpenCV + MediaPipe real-time pupil tracking, 5-level severity classification. Paper authored, advancing toward clinical trials.

PythonOpenCV MediaPipeVercel
02

Vehicle Classifier

Problem: Manual traffic monitoring is slow and error-prone in high-density urban environments.

YOLOv8 pipeline classifying 5 vehicle types from live feeds with REST API reporting. Built at Samsung R&D Institute, Kolkata.

Production-grade CV pipeline detecting two-wheelers, sedans, SUVs, buses and trucks — engineered inside a live R&D environment with professional engineers.

PythonYOLOv8 Deep LearningREST API
03

Market Sentiment Analyser

Problem: Manually reading thousands of customer reviews is a bottleneck for product teams.

NLP pipeline classifying reviews as positive/negative/neutral with real-time visualisations and CSV batch upload.

HuggingFace Transformers + VADER pipeline with interactive Plotly dashboards. Instant polarity analysis at any scale.

PythonNLP HuggingFacePlotly
04

Sales Lead Scorer

Problem: Sales teams waste time chasing low-probability leads without data-driven prioritisation.

GBM classifier scoring leads 0-100%. Real-world inspiration: 12% ROI improvement, Rs. 2.5L+ incremental revenue at MyGlamm.

Trained on 1,200 synthetic leads — full pipeline analytics and model explainability dashboard with feature importance visualisation.

PythonScikit-learn GBMData Science

"Precision over speed. Always."

The method.

01
◎
UNDERSTAND

Identify the real problem before writing a single line of code. Most failures are diagnosis failures.

→
02
⬡
ENGINEER

Build with precision — every architectural decision is intentional, every trade-off is considered.

→
03
◈
VALIDATE

Test against real data. Measure outcomes, not assumptions. r=0.9857 or it ships again.

→
04
↗
SHIP

Deploy, monitor, iterate. The work is never truly done — only the version number changes.

"The path is made by walking."

The chapters.

Aug 2023
The Beginning

Enrolled in B.Tech CSE at Brainware University, Kolkata. Currently maintaining 8.79 SGPA.

Jun 2023 – Aug 2024
Marketing Coordinator — MyGlamm

Market research across 50+ stores. 12% campaign ROI improvement, 18% customer acquisition rise, Rs. 2.5L+ incremental revenue.

Sept – Nov 2025
Samsung R&D Institute — AI & API Integration

Delivered production-grade Vehicle Classifier pipeline integrating YOLOv8 with REST APIs in a live engineering environment.

2024 – 2025
4 AI/ML Projects Shipped

Built and deployed Ocular AI, Vehicle Classifier, Market Sentiment Analyser and Sales Lead Scorer — all live.

Feb 2026
ISI Kolkata — Cryptographic Security

5-day MeitY-funded intensive on Privacy Preserving Cryptography, Advanced Encryption & Zero-Knowledge Proofs.

Mar – Apr 2026
Qualcomm · Microsoft · SEBI Certifications

Qualcomm Deep Tech Entrepreneurship — 100% score. Microsoft × Simplilearn MS Excel. SEBI Financial Literacy Quiz 2026.

Jun 2026
IGNIS Launched · Ocular AI v2

Launched IGNIS — full-stack personalised fitness platform on Next.js + Supabase + Claude AI. Upgraded Ocular AI to v2 on Vercel.

Jul 2027 →
Final Year. Open to Opportunities.

Completing B.Tech CSE. Actively seeking AI/ML roles, research collaborations, and ambitious problems worth solving.

"Tools are only as sharp as the mind wielding them."

What I wield.

⬡
Python
Primary Language
Core
▲
Next.js · React
TypeScript · Tailwind CSS
◎
Supabase
Database · Auth · Backend
👁
Computer Vision & Deep Learning
OpenCV · MediaPipe · YOLOv8 · TensorFlow · PyTorch
◈
NLP
HuggingFace · Transformers · VADER
▦
Data Science
Pandas · NumPy · Plotly · SQL
🔐
Cryptography
Zero-Knowledge Proofs · Encryption
ISI · MeitY
↗
Deployment
Vercel · Streamlit · GitHub Pages
⑂
Git & GitHub
Version Control · API · Automation

"The best collaborations start with a single message."

Let's build
something.

Open to AI/ML roles, research collaborations, freelance projects, and conversations worth having.