Full Stack AI Engineer building production-ready applications. Bridging sleek interfaces, resilient backends, and ML workflows to deliver complete end-to-end products.
I work across the whole product stack: interface, backend, data flow, model integration, evaluation, and automation.
Full Stack AI Engineer based in India, bridging model prototyping with enterprise deployment. I engineer responsive frontends, scalable API backends, and modular AI pipelines.
Specialized in Python, RAG systems, ML, and agentic workflows, focusing on low-latency performance, clean architecture, and reliable execution.
At Ellucian, I worked on applied AI systems that turn documents, transcripts, and knowledge workflows into usable automation.
Ellucian
Integrated AI workflows into Ellucian's enterprise ecosystem, transforming raw documents, meeting logs, and databases into automated solutions.
Large, technical builds presented as case-study ready product surfaces, with clear problem framing and credible implementation signals.
Problem: Standard steganography triggers detection anomalies when hiding large payloads.
Solution: Built a dual-carrier deep learning system with RSA-2048 + AES-256 encryption, embedding payloads across image and text structures.
Impact: Achieved sub-perceptible distortion (PSNR > 42dB) and full resistance to steganalysis scans.
Problem: Online resume and prep tools are slow and compromise user data privacy.
Solution: Built an offline-first job suite with local PydanticAI agents for CV tailoring, mock interviews, and DSA prep.
Impact: 100% data privacy and zero API costs via local client inference.
Problem: Standard plagiarism tools miss paraphrased and machine-translated text.
Solution: Designed a cross-encoder semantic pipeline analyzing word embeddings and stylistic perplexity.
Impact: 94% detection precision with detailed integrity reports in under 4s.
Problem: Support teams waste hours parsing complex ERP technical docs.
Solution: Built a RAG engine with ChromaDB vector search and instant acronym resolution.
Impact: Cut manual doc lookups by 35% with automatic citation routing.
Problem: Identifying Congenital Heart Disease from X-rays has high diagnostic variability.
Solution: Constructed an ensemble system combining InceptionV3 with SVM and Random Forest classifiers.
Impact: 89.4% clinical accuracy, reducing diagnostic error flags by 18%.
Problem: Cold-start data issues cause inaccurate book recommendations.
Solution: Built a recommendation engine using TF-IDF and cosine similarity on plot synopses.
Impact: Sub-100ms response latency and 1.8x increase in user engagement.
Organized by delivery capability instead of raw tags, with a focus on practical production workflows.
Interfaces and interaction
Models, retrieval, evaluation
APIs and services
Storage, deployment, workflow
I am open to full stack AI engineering roles, AI/ML internships, product collaborations, and projects that need a strong bridge between software engineering and applied intelligence.