Headline

Software Engineer at CogniQuest AI || Gen AI || Python || AI/ML

About

I'm a Software Engineer with 3+ years of experience designing and shipping Generative AI systems for enterprise document intelligence. I focus on building production-grade GenAI pipelines—combining LLMs, RAG architectures, and hybrid retrieval (BM25 + vector search) with OCR and computer vision for structured data extraction. I care about the parts that are easy to skip: prompt engineering that actually holds up at scale, retrieval that stays accurate under messy real-world data, and systems that don't fall over under load. Some things I'm proud of: 🔹 Built GenAI pipelines achieving 85% extraction accuracy on unstructured documents 🔹 Cut manual effort by 60% across e-commerce document workflows 🔹 Scaled systems to process 10,000+ documents/day in production On the backend side, I work primarily in Python, FastAPI, and Celery, designing distributed microservices with an emphasis on performance and reliability — from initial architecture through to production deployment and beyond. I enjoy the full lifecycle: framing the problem, prototyping with LLMs, hardening the pipeline, and then obsessing over the metrics that tell you whether it's actually working. Open to connecting with folks working on applied GenAI, document AI, or retrieval systems — always happy to trade notes.

Location

Bengaluru

AI stack

anthropic apis

claude

gemini

hugging face

langgraph

mcp

Work experience

Software Engineer at Cogniquest AI, Jan 2024 - Present · 2 yrs 11 mos, Bengaluru, Karnataka, India, • Led development of a distributed AI-driven invoice processing system for a large-scale e-commerce enterprise, handling large-scale document workflows across thousands of invoices daily. • Architected end-to-end pipelines extracting structured financial data (GSTIN, invoice values, metadata) using LLMs, OCR, and NLP models, achieving 85% extraction accuracy. • Reduced manual validation effort by 60% through hybrid ML + rule-based approaches, saving high operational costs for the client. • Implemented hybrid retrieval (BM25 + vector search) with Llama-based models for improved contextual understanding across document sets. • Designed and built table parsing and financial computation engines to validate tax, taxable values, and totals with high precision. • Built a cross-document validation system to reconcile invoice data against supporting documents, ensuring data consistency at scale. • Engineered CV-based extraction pipelines using bounding box analysis and sign detection for precise field localization. • Designed async processing pipelines (Celery + Redis) handling 10,000+ documents per day with sub-2s average processing latency. • Developed Dockerized backend services for scalable document processing systems, reducing deployment time by standardising environments. • Built pipelines to transform unstructured data into structured formats for downstream ML workflows. • Contributed to AI-based data extraction and automation workflows adopted in the production invoice intelligence platform.

Software Engineer at BrainyBeam Technologies Pvt. Ltd., Jul 2023 - Aug 2023 · 2 mos, Ahmedabad, Gujarat, India, • Developed backend services using Python and Django, improving API response time by 20% through query optimization and code refactoring. • Gained hands-on exposure to REST API design, database schema modelling, and agile development workflows.

Education

Bachelor of Engineering - BE, Computer Engineering, VADODARA INSTITUTE OF ENGG., KOTAMBI, WAGHODIYA 080

Science, Shannen School