👋 Hi, I'm Aaditya Punekar
I build Data|
that drives decisions.
Data Scientist & AI Engineer with expertise in Machine Learning, NLP, and GenAI. Building predictive models and intelligent automation that deliver real business value.
# AI-powered prediction engine import torch, transformers from sklearn.ensemble import XGBClassifier class RetailBrain: def predict(self, data): forecast = self.model(data) return { "stockout_risk": forecast.prob, "confidence": 0.94 }
About Me
Empowering businesses with intelligent data solutions.
I am a Data Scientist & AI Engineer with over 2 years of experience applying machine learning, NLP, and GenAI tools to deliver actionable insights and automation across retail, pharma, and insurance sectors.
With a proven track record in building predictive models, streamlining data workflows, and driving data-informed business decisions, my goal is to bridge complex algorithms with tangible value. I specialize in deploying LLMs, constructing dashboards, and scaling data architectures using Python, SQL, and modern cloud deployment strategies.
Currently based in Mumbai, India, I hold an MSc in Computer Science from Queen Mary University of London.
Experience
Skills & Technologies
Languages & Frameworks
AI / ML
Data & Cloud
Experience & Education
Trading Manager
J Sainsbury's · London, United Kingdom
Built custom Power BI reports and queried store transactional data using SQL to track daily store P&L, sales trends, and labor efficiency, directly enabling decisions that exceeded target budgets by 7.1%. Analyzed sales declines and inventory waste variances via SQL, designing automated markdown timing workflows that minimized stock loss. Developed footfall and stock consumption dashboards for high-volume Wembley event days (sales surges up to £60k/day). Maintained 100% shrink driver compliance against a 95% regional benchmark.
Data Analyst
Deep Punch · Remote, Mumbai
Analyzed supplier performance metrics and material cost trends using Python and Excel, driving a 15% reduction in overall purchasing costs. Deployed real-time sentiment analysis and Named Entity Recognition (NER) models using Python and Hugging Face Transformers, reducing processing latency by 35%. Built custom Power BI dashboards for sourcing and finance teams, supporting strategic contract renegotiations.
MSc Computer Science
Queen Mary University of London
Specialised in advanced computing algorithms and data science paradigms. Graduated with a focus on machine learning systems and intelligent data architectures.
BE Computer Engineering
University of Mumbai
Built a strong engineering foundation in software architecture, computational logic, and data structures.
Featured Projects
Retail Brain — Sainsbury's
Full-stack AI demand forecasting system for Sainsbury's. Features JWT-authenticated REST API, real-time WebSocket alerts, POS connectors, and ML-powered stockout prediction for 1,000+ SKUs with 94% accuracy.
Retail Ops Copilot
End-to-end decision intelligence system using Reinforcement Learning to automate staff task assignments. Features MILP optimization baseline, digital twin simulation, and an interactive dashboard for retail operations.
AgentSentinel
Multi-agent AI monitoring system providing real-time oversight of autonomous AI pipelines. Implements structured guardrail layers, audit logging, and enterprise-grade observability for agentic AI workflows.
YouTube Automation Pipeline
Fully automated AI-driven content pipeline that generates scripts, voiceovers, thumbnails, and video compilations using GPT-4 and ElevenLabs. Targets fitness & health niche with engaging AI personas.
Logistics Reconciliation Engine
Zero-hallucination AI agent for retail logistics that eliminates "Ghost Inventory" by autonomously ingesting unstructured supplier data (PDFs, emails) and reconciling against master inventory with structured discrepancy reports.
GenAI Mention Intelligence
LLM-powered brand and entity mention analysis system. Fine-tuned Longformer models for multi-label entity tagging in pharma and insurance datasets, improving classification accuracy by 30–40% with automated JSON pipelines.
Excel to PDF Summarizer
AI-powered document intelligence tool that ingests Excel and CSV reports, applies LLM-generated natural language summaries, and exports polished PDF reports. Reduces manual reporting time significantly.
Hybrid Malware Detection
High-precision malware classification system using PE header analysis and ensemble models. Engineered feature extraction pipelines combining CNN binary visualisation, XGBoost, and SVM architectures.
Let's Collaborate
Let's build something
intelligent together.
I'm open to new opportunities in data science, ML engineering, and AI product development. Whether it's a full-time role, contract work, or a collaboration — let's talk.