Bharat Singh
AI Systems Architect · AI Engineering Lead
Professional Summary
AI Systems Architect and Engineering Lead with 20+ years of experience directing high-throughput Information Retrieval (IR) architectures, autonomous agent platforms, distributed systems, and knowledge graphs across European institutions and enterprise SaaS. Proven track record leading engineering squads (8–9 developers + QA), executing high-risk core engine modernizations with ~99% production parity, and engineering multi-tenant RAG and search pipelines scaling across 3,500+ commercial websites under sub-50ms p95 latency SLAs.
Technical Expertise
Agentic AI & Orchestration: LangGraph, Model Context Protocol (FastMCP / SSE), Stateful Multi-Agent Loops, Tool Execution & Function Calling, Supervisor-Worker Pipelines, Deterministic Guardrails.
Information Retrieval & RAG: Multi-Tenant Search Platforms, Hybrid Search (Dense Vector + BM25), Cross-Encoder Neural Rerankers, Reciprocal Rank Fusion (RRF), Local Embedders (BGE, FastEmbed, ONNX Runtime), Quantization (int8, FP16), Vector Stores (MongoDB Atlas Search, Qdrant, pgvector).
Systems & Platform Architecture: Hexagonal Architecture, Protocol-Based Interfaces, High-Throughput Rendering Engines, Distributed Workflow Orchestration (Prefect 3.x), Microservices (FastAPI, gRPC, REST), Parity Verification (~99% Production Parity).
Infrastructure & Operations: Kubernetes (AKS), Docker Containerization, Azure DevOps CI/CD Pipelines, Azure Blob Storage, Linux System Internals, LangSmith Tracing, Telemetry.
Languages & Databases: Python (FastAPI, Pydantic, Prefect, PyTorch), Java, Rust, SQL, MongoDB, Neo4j, GraphDB, RDF/SPARQL.
Professional Experience
AI Application Tech Lead (Full-Time)
· Milestone Internet
Sep 2025 – Present · Ahmedabad / Remote (India)
- Core Platform Modernization: Led the architectural re-engineering of the enterprise content serving engine to asynchronous Python on Kubernetes; achieved ~99% golden rendering parity across 20+ production pods and 3,500+ customer domains, delivering ~$70,000/year in projected infrastructure cost savings.
- Autonomous Multi-Agent Platform: Directed cross-functional engineering squad (8–9 developers + QA) building enterprise multi-agent workflows; designed stateful cyclic graph execution and decoupled tool contracts (FastMCP over SSE/REST), eliminating conversational state loss across complex multi-turn booking and support journeys.
- In-House Neural Inference & Hybrid Search: Built and deployed dedicated local inference microservices for neural embeddings and cross-encoder rerankers, replacing third-party API dependencies and achieving sub-50ms p95 retrieval latency SLAs under peak traffic.
- AI Governance & Automated Telemetry: Instituted enterprise-grade compliance and observability platform incorporating automated PII scrubbing (HIPAA/GDPR compliance) and unsupervised semantic drift clustering over user query streams to identify failure modes proactively.
Enterprise AI Consultant (Contract)
· Milestone Internet
Jun 2025 – Sep 2025 · Remote (India)
- Multi-Tenant Ingestion & Search Platform: Architected the foundational multi-tenant document ingestion and vector search platform powering 3,500+ commercial websites; designed resilient DOM scraping, semantic chunking boundaries, and vector indexing pipelines to ingest heterogeneous web catalogs with high availability.
- Engineered configurable discovery, parsing, chunking, and indexing stages, establishing the core retrieval infrastructure for customer experience applications.
AI Specialist Consultant (Contract)
· European Patent Office (EPO)
2023 – Jan 2025 · The Hague / Remote (India)
- Architected multilingual information extraction and semantic search pipelines indexing millions of international patent records.
- Applied Word Sense Disambiguation (WSD) and concept indexing to improve prior-art detection recall and claim classification accuracy.
- Engineered NLP summarization and clustering tools for internal patent examiner tooling.
Knowledge Graph Consultant (Contract)
· Knewco Inc.
2021 – 2023 · Rockville, MD / Remote (India)
- Developed enterprise knowledge graphs using unsupervised relation extraction and semantic inference engines.
- Formulated ontology alignment and concept resolution strategies across large heterogeneous scientific datasets.
AI & Data Modeling Lead (Contract)
· EU COVID Task Force
2020 – 2021 · Brussels / Remote (India)
- Led cross-national health and mobility data schema modeling and data integration pipelines across EU member states.
- Developed interpretable epidemiological machine learning models and visualization pipelines for international policy reporting.
AI Consultant & Scientific Researcher (Contract)
· Medical Informatics Research Consortia / NBIC
2014 – 2020 · Netherlands / Remote (India)
- Advised on semantic knowledge representation, graph databases (RDF, SPARQL), and clinical text mining engines.
- Pioneered graph-based lexical semantic disambiguation models across biomedical ontologies (UMLS, MeSH).
Senior Software Engineer & Scientific Researcher
· Erasmus Medical Centre, Rotterdam
2010 – 2014 · Netherlands / Remote (India)
- Engineered full-stack semantic web architectures, graph stores, and natural language processing pipelines.
- Co-authored peer-reviewed research on clinical entity normalization and data value modeling.
Senior Software Engineer (Full-Time)
· Leiden University Medical Centre / NBIC
2008 – 2010 · Leiden, Netherlands
- Developed semantic web data layers and graph databases integrated with enterprise Java services.
- Built unsupervised text processing pipelines for clinical concept normalization.
Senior Software Engineer (Full-Time)
· Quark Inc.
2004 – 2008 · Chandigarh, India
- Engineered enterprise-scale publishing platforms and high-performance document processing engines in Java.
Education
Master of Computer Applications (MCA)
· Punjab Technical University
India
Peer-Reviewed Publications
The value of data: Nature Genetics, 43(4), 281–283 (2011). doi:10.1038/ng0411-281
Contribution: Contributed to semantic data modeling and nanopublication specification implementation for machine-readable assertion networks across scientific corpora.
Contribution: Contributed to semantic data modeling and nanopublication specification implementation for machine-readable assertion networks across scientific corpora.
Using rule-based natural language processing to improve disease normalization in biomedical text: JAMIA, 20(5), 876–881 (2013). doi:10.1136/amiajnl-2012-001173
Contribution: Designed and implemented rule-based NLP pipelines and dictionary lookup algorithms for normalizing clinical disease mentions against UMLS and MeSH ontologies.
Contribution: Designed and implemented rule-based NLP pipelines and dictionary lookup algorithms for normalizing clinical disease mentions against UMLS and MeSH ontologies.
Using an ensemble system to improve concept extraction from clinical records: Journal of Biomedical Informatics, 45(3), 423–428 (2012). doi:10.1016/j.jbi.2012.01.001
Contribution: Developed ensemble concept extraction pipelines combining machine-learning classifiers with rule-based heuristics to extract clinical entities from unstructured records.
Contribution: Developed ensemble concept extraction pipelines combining machine-learning classifiers with rule-based heuristics to extract clinical entities from unstructured records.