From writing my first web crawler in 2006 to architecting enterprise GenAI platforms today — every role taught me something the next one demanded. This is the full story.
Eterno Infotech (DailyHunt / Newshunt)
Started my career building the backend for DailyHunt — then called Newshunt — India's largest regional news aggregator. Designed and implemented a web crawler ecosystem that consumed content from 80+ newspaper publishers in formats ranging from rich XML and HTML to JSON feeds. Built the content ingestion pipeline, deduplication logic, and a metadata tagging engine that powered the app's recommendation layer.
This was where I learned that software at scale is fundamentally different from software in theory. Millions of mobile users depended on that pipeline running without a hiccup, and that lesson — reliability under pressure — shaped everything that followed.
Impetus Technologies (Client: American Express)
The pivot point. Moved from traditional software engineering into the Big Data world. Led the development of SERT — the Speed, Engagement, and Relevance Tool — for American Express's open-card merchant ecosystem. Designed Hadoop-based ETL pipelines that processed terabytes of transactional data for merchant recommendations and engagement scoring.
Built recommendation engines using collaborative filtering on HBase. Introduced Elasticsearch for real-time search across millions of merchant profiles. This is where I discovered that data, not code, is the real product — the code just shapes it. The pivot from writing applications to designing data architectures was the most transformative shift of my career.
Inkers Technologies
Took the leap into Deep Learning. Led a team building computer vision solutions for retail analytics — face recognition systems for customer identification, crowd counting algorithms for footfall analysis, and age/gender estimation models for demographic profiling.
This was hands-on, research-to-production work: training custom CNNs, implementing face embedding with triplet loss, optimizing inference for real-time on commodity hardware. Simultaneously pursued Udacity's Self-Driving Car and NLP nano degrees, building the theoretical foundation that would power the next phase.
Trigyn Technologies
Scaled computer vision from single-camera demos to an enterprise platform: TIVA (Trigyn Intelligent Video Analytics). Led end-to-end development for real-time analytics across 1000s of video streams — object detection, intrusion alerts, people counting, ANPR, and anomaly detection — deployed on Jetson Nano/TX2 edge devices.
Managed a multi-skill team of 12 spanning embedded engineers, backend developers, frontend designers, and data scientists. Introduced model pruning and quantization strategies that cut false alarms by 75% while reducing deployment costs 17-20%. This role taught me that AI leadership is equal parts technical depth and people orchestration.
Hexad GmbH (Client: Volkswagen)
The bridge between Deep Learning and Generative AI. At Volkswagen, architected two critical projects. legalAI — a document intelligence platform that parsed complex legal PDFs using CV-based layout analysis combined with semantic merging and entity logic, doubling throughput and improving extraction accuracy by 37%.
Then QAM — a Question Answering Machine for VW's hardware knowledge base, built on FLAN-T5 with VectorDB retrieval. This was my first production RAG system, and it revealed the immense potential of combining retrieval with generation. The NLP Architect role solidified my understanding that the future wasn't just CV or NLP — it was multimodal intelligence.
Cigna — AI Center of Excellence
Designing the architecture for Cigna's enterprise GenAI pipeline — Project Genius. Building prompt + few-shot retrieval systems using custom fine-tuned models (embedding + CNN) for complex table extraction from healthcare documents. The pipeline integrates FAISS for vector search, LLM guardrails for safety, PII/data controls for HIPAA compliance, and evaluation gates for quality assurance.
This role brings together everything: the engineering discipline from Era 01, the data architecture from Era 02, the model optimization from Era 03, and the NLP/retrieval expertise from the VW transition. Serving as ML Advisor to the AI CoE, establishing architecture standards, cost/latency SLOs, and security-by-design patterns for the organization's GenAI roadmap.
Simultaneously completing my M.Tech in AI & Machine Learning from BITS Pilani (2023-2025), deepening the theoretical foundations in Deep Neural Networks, Reinforcement Learning, NLP, and Information Retrieval.
Every era was fueled by deliberate learning — formal degrees, nano degrees, and 14+ certifications
BITS Pilani, India
Deep Neural Networks, Deep Reinforcement Learning, NLP, Statistics & Probability, Information Retrieval — pursued while working full-time at Cigna
VTU, India — First Class
Operating Systems, Computer Networks, Parallel Programming, Algorithms & Data Structures — the engineering fundamentals
Vehicle Detection & Tracking, Traffic Sign Classification, Behavioral Cloning, Sensor Fusion, Path Planning
PoS Tagging (HMM), Machine Translation (DNN), Speech Recognition (RNN/CTC)
Generative AI with LLMs, Neural Networks & DL, CNNs, Sequence Models, AI for Medical Prognosis & Diagnosis, Hyperparameter Tuning, and more
Sun Certified Java Programmer & Sun Certified Web Component Developer — the original era credentials
The next era is being written. I'm exploring agentic AI, multi-modal reasoning, and scaling GenAI for regulated industries. Want to build the future together?