Jun 2023 – Nov 2023 · 5 mos

• For brief period of time was managing the projects, a lesson learnt on job.

Apr 2022 – Present · 4 yrs 1 mo

• 🔹[Samsung TV Plus (FAST) | Real-time Discovery (RecSys + Search/IR)]
• Built p95 low-latency real-time personalization + discovery stack for remote-first TV UX across ~4,300 live channels + ~66,000 VOD in ~30 countries.
• Unified RecSys + Search/IR across Guide, Home, VOD rails, Sports hubs with hybrid retrieval (ANN embeddings + lexical + rules) and freshness + geo/rights + language/safety filters.
• Implemented constraint-aware ranking for channel-surfing intent, cold-start, multilingual catalogs, and diversity/novelty; deployed distributed serving with cache hierarchy (device → edge → server) + online features.
• Primary metrics owned: Time-to-First-Play (TTFP), Qualified Watch Time, Retention, Query Success Rate, Latency/SLO (p95), Infra cost per play, user satisfaction.
• Revenue linkage: improved discovery → higher watch time/retention → better ad fill + ad yield (FAST monetization) and session depth (measured via qualified watch-time uplift + retention lift).
• 🔹[Samsung.com (E-Com B2C) | Batch Product Personalization ($120M Revenue)]
• Architected a daily batch recommender across 20 countries for Homepage, PDP, Search, Order Confirmation, using 6+ algorithms (EASE, SASRec, CF, FastText similarity, association rules, popularity).
• Built BigQuery pipelines (Adobe clickstream + orders + catalog) with daily feature refresh; trained EASE + FastText on Dataproc; orchestrated DEV/STG/PROD with Airflow (retries/backfills/logging).
• Developed a modular framework (40+ strategies) with blending + business rules + country tuning; shipped GDPR-aligned personalization (consent filtering, governed PII).
• Business impact: drove $120M cumulative revenue from personalized placements and improved conversion across key surfaces.
• Primary metrics owned: CTR, ATC rate, CVR, Revenue per Session / RPV, AOV uplift, coverage (catalog + user), diversity, latency/availability of daily delivery, and offline ranking metrics (NDCG/Hit@K) for model iteration.

Jun 2023 – Present · 2 yrs 11 mos · Remote

• Organize Events for AICamp and Plan the Host Setup. Assisting Tech Crowds with AI Related Questions

Apr 2021 – Present · 5 yrs 1 mo

Jul 2020 – Present · 5 yrs 10 mos

• Test courses from Coursera and Suggest on the content before they are released; and give my reviews and feedback on the content.Primarily Data Science and Analytics

May 2019 – Present · 7 yrs

• Kaggler by Curiosity , 3X Kaggle Expert

Feb 2019 – Mar 2022 · 3 yrs 1 mo · Remote

• [Domain: Claim Estimation in Automotive & Property Insurance]
• 🔹Worked on creating fully supervised and semi-supervised semantic segmentation models to segment different types of damage on images of damaged cars.
• 🔹Incorporate Computer Vision, AI, 3D rendering, etc. into existing product for Vehicle Claim
• Verification by creating synthetic dataset by Unity3D, UnReal Engine to train model.
• 🔹Developed pipelines for data preprocessing and image feature extraction and reduced reproduction GPU hours and cost by 50%.
• Worked with CCCOne Total Repair Platform to integrate APIs.
• For Property Inspection Tool worked to identify Start-Up Tech and Evaluation for Allstate Strategic Ventures.
• Technical Lead
• 🔹Allstate a Fortune 100 Company (Closely worked with Allstate Northern Ireland Team)

Oct 2015 – Jan 2019 · 3 yrs 3 mos · Remote

• Worked extensively in Computer Vision , Image Processing and Edge AI based product development which includes Algorithm Experimentation, Implementation, Optimization and Operationalization.
• Below are the Work in brief :
• Scene Text Detection &Recognition : Deep Learning based Detection, Extraction , Recognition and Translation . [Domain: Logistics & Transportation].
• Digital Evidence Management System: Object Detection & Localization, Tracking the object
• Masking portion of image. [Domain: surveillance]. [Live-In Production]
• Forest Asset Inspection : Identify tree , Localization, counting of trees , Volumetric Analysis using 3D Point Cloud and Digital Terrain Model. [Domain :Forestry].
• Warehouse Inventory Management: Counting the inventory in warehouse, Image Quality Analysis, Image Alignment and Super Image Resolution. [Domain: Retail & Warehouse]. [Live-In Production]

Oct 2014 – Sep 2015 · 11 mos · Hybrid

• Worked involved Ontology Modelling and Development on Tenant Management System
• Used RDFS, RDF, writing SPARQL Query ,using Sesame API & Jena API.
• Triple Store in Open Source Technologies. [Domain: IOT]
• CMC earlier a PSU later acquired by TCS Ltd in 2014

Apr 2014 – May 2014 · 1 mo · online

• My Responsibility as an Intern was to generate traffic and encourage the awareness about Car Pooling ,Car Sharing to avoid traffic congestion.