Siddartha A Y
AI Systems • Backend • Cloud Infrastructure • Applied ML / GenAI
From production AI for Airbus & Boeing to founding-engineer roles at US startups — I build high-reliability AI, backend, and distributed systems optimized for correctness, throughput, and scale.
Final-year CS student at PES University. I've shipped safety-critical LLM pipelines for Airbus and Boeing, designed the entire AI services architecture for a US-based EdTech startup as a founding engineer, and helped create India's largest public urban traffic dataset at IISc Bangalore — all before graduating.
I work at the intersection of AI systems, backend engineering, and cloud infrastructure, building LLM pipelines, scalable services, containerized deployments, and DevOps workflows that handle real load, fail gracefully, and remain maintainable in production.
Sole AI engineer for a DO-178C/ARP4754A-aligned aerospace traceability engine, architecting a multi-stage lexical+dense retrieval, cross-encoder reranking, and LLM obligation-decomposition pipeline over hierarchical requirement graphs for certification-grade verification.
Designed a production-grade agentic inference and memory system for a US EdTech startup, integrating schema-constrained 2-pass LLM orchestration, O(1) Redis I/O, async cache hydration, and token-efficient session compression.
Contributed to UVH-26, India's first large-scale open-source urban traffic dataset — fine-tuned six SOTA detection models and achieved up to 31.5% mAP improvement over COCO baselines.
Designed and implemented RACD — a Retrieval-Augmented Change Detection architecture that outperforms pixel-differencing baselines with queryable external exemplar memory at inference time.
Designed and deployed a production RAG pipeline on AWS powering an LLM recruitment engine — end-to-end: model training → containerised microservices → cloud infra, cutting screening overhead by ~30%.
Chairman of IEEE CS PES University chapter — directing technical strategy across CV, GenAI, and Deep Learning verticals for one of PES University's largest technical bodies.
Tech Lead of NeuralHive, the official AI/ML club at PES University — running workshops, hackathons, and applied ML initiatives.
Awarded Best Breakthrough Idea at Google Agentic AI Day 2025 — selected from 1,000+ competing teams across India.
Top 8 finalist out of 1,000+ teams in a national-level product and engineering challenge.
First Runners-Up at a national-level hackathon.