AI Architect & Forward Deployed Engineer · Pune, India

7+ years building cloud-native, distributed systems in Go, Python, and TypeScript on GCP, AWS, and Azure. I specialize in Agentic AI systems and multi-agent orchestration, AI governance and compliance, healthcare AI, microservices, gRPC / GraphQL APIs, and FinOps — bridging LLMs with real-time enterprise systems across FinTech, healthcare, and energy (Google ADK, MCP, Microsoft Agent Framework, Multi-Agent Reference Architecture, Vertex AI Agent Engine, LangChain, LangGraph, CrewAI).

Merged correctness, concurrency, and security fixes into the official agent frameworks of Microsoft and Google and the Model Context Protocol standard — plus first merges into official Apple and AWS repos — the core building blocks of the AI-agent ecosystem (20k+ combined stars). Not demos or forks: code landed in main, each with a regression test.

Microsoft's official Go implementation of the Agent Framework. My 234 merged PRs span runtime correctness across the Anthropic and OpenAI providers, the workflow engine, A2A, and MCP layers — each shipped with a regression test. Selected highlights:

• Closed leaked streaming HTTP bodies on the Anthropic / OpenAI-Responses paths — #718
• Released dead references in concurrent.Queue.Dequeue — #598
• Fixed empty responses in the structured-output middleware — #607
• Fixed a nil-update panic in the tool-approval middleware — #472
• Removed a dead PortableValue guard in the workflow route builder — #489

Google's official Agent Development Kit for Go. Runtime-correctness fixes of the same character as my Microsoft work:

• Guarded a nil first candidate in the streaming response aggregator — #1802
• Honored MaxConcurrentRuns in the Pub/Sub controller — #1720
• Maintained session-state UpdateTime on app/user state writes — #1204

The official Go SDK for the Model Context Protocol — the industry standard for connecting AI agents to tools and data.

• Prevented a MemoryEventStore.After panic on an index past the end — #1132
• Fixed the JSON-RPC response summary to use %v and ID.Raw() — #1382

A second official Microsoft repo — agent security and governance. Merged security fixes, with further hardening in review:

• Enforced the configured sandbox execution timeout in the Docker provider — #3567
• Added opt-in PII redaction to the credential redactor — #3565

Correctness and crash fixes across the schema conversion pipeline of Google Cloud's official Spanner migration tool:

• Preserved the UNIQUE flag when converting GoogleSQL indexes — #1422
• Fixed a nil-pointer panic on nextval() with a bare string argument — #1467
• Keyed ALTER TABLE unique indexes by column ID instead of name — #1426

• Consulting — forward deployed engineering for AI teams: production readiness audits, POC-to-production rescues, and fractional embedded engagements
• AI transformation for NESO (National Energy System Operator) — enterprise AI adoption with a strong focus on AI governance and compliance
• Multi-agent AI systems for FinTech (Google ADK · MCP · Multi-Agent Reference Architecture · Vertex AI Agent Engine)
• Medical AI — FHIR R4 + HL7 v2, cost-aware diagnostic budgets, clinical reasoning
• RAG & GraphRAG pipelines on AlloyDB AI, BigQuery Knowledge Graph, and pgvector
• Voice AI assistants (ElevenLabs + Gemini)
• Open Banking — consent management and regulatory compliance (ADGM, DIFC, SAMA)
• Open-source contributions — Microsoft Agent Framework Go, Genie, Bodh, and Google's Spanner Migration Tool

Consulting engagements (fractional forward deployed engineer, audits, POC rescues — services) · AI Architect, Staff Engineer, and Solution Architect roles in Agentic AI, FinTech, Healthcare Tech, GenAI Infrastructure, and Open Banking — Bengaluru, Pune, Mumbai, Hyderabad, Remote, EMEA, UAE.

• ✉️ Email — i.pratikdhanave@gmail.com
• 💼 LinkedIn — linkedin.com/in/pratikdhanave (2,750+ followers)
• 🌐 Portfolio — pratikdhanave.com
• 📅 Book a call — calendar.app.google/x7W5bAdst4J6MT2u7 · topmate.io/pratik_dhanave
• 🎤 Google Cloud Next 2022 Talk — Migrating Monolith Applications into Microservices

I publish in-depth, multi-part technical series at pratikdhanave.com/blog — 118+ series (1,320+ posts) across agentic AI, distributed systems, and production engineering, each grounded in primary sources with runnable examples. Browse all series →

Featured series

• The AI Production Roadmap — the 12-phase path from prototype to production AI: governance, evaluation, serving, security, cost
• Distributed Systems from First Principles — consistency models, CAP/PACELC, replication, consensus (Raft), resilience
• LLM Inference & Serving — the KV cache, continuous batching, quantization, speculative decoding, vLLM/PagedAttention
• The AI Cost Optimization Playbook — the ranked inference cost levers, vendor-grounded and measured
• Adopting AI in the Enterprise — the organizational playbook for crossing pilot-to-production: use cases, readiness, agentic orchestration, scale, security, change
• Agent Memory Systems — how agents remember, learn, and forget: the memory hierarchy, compaction, retrieval, consolidation, production memory

Deep dives

• Multi-Agent Systems Architecture — hierarchical coordination patterns, function calling, cost-aware decision trees, observability & cost tracking https://pratikdhanave.com/blog/posts/adk-03-multi-agent-systems.html
• BigQuery FinOps: 57% Cost Reduction at Scale — query refactoring, materialized views, MERGE optimization, slot-based capacity planning https://pratikdhanave.com/blog/posts/57-percent-bigquery-cost-reduction-tata.html
• FHIR R4 + HL7 v2 Medical AI Interoperability — Bodh architecture, cost-aware diagnostic budgets, HIPAA compliance patterns https://pratikdhanave.com/blog/posts/hl7v2-still-matters.html

• 🎤 Speaker at Google Cloud Next 2022 — "Migrating Monolith Applications into Microservices"
• 🌍 Google Summer of Code — Contributor (2017–2018), Mentor (2019–2026); guided 10+ students on cloud-native and AI-native projects
• 💼 Productized Tata Group BigQuery cost-optimization engagement into Searce's GCP managed service offering — INR ₹100 Cr+ (~$12M) saved, 57% DW cost reduction
• 🎓 Google Cloud Generative AI Leader Professional Certificate
• 🎓 Google Cloud Specialization — Data Engineering, Big Data, and Machine Learning on GCP
• 🎓 Ardan Labs — Ultimate Go and Ultimate Services Go

• Microsoft Agent Framework Go — 234 merged PRs into Microsoft's official Go Agent Framework (672★). See the featured contributions above.
• Google ADK (Go) — 17 merged PRs into Google's official Agent Development Kit for Go (8.9k★) — nil-guards, concurrency, and session-state correctness.
• MCP Go SDK — 8 merged PRs into the official Go SDK for the Model Context Protocol (5.3k★) — panic prevention and JSON-RPC correctness.
• Microsoft Agent Governance Toolkit — merged agent-security fixes (sandbox timeout enforcement, PII redaction) into a second official Microsoft repo (6.5k★).
• Google Cloud Spanner Migration Tool — 12 upstream fixes in review across the schema-conversion pipeline of Google Cloud's official migration tool: panic guards, GoogleSQL index correctness, error-message accuracy.
• GoogleCloudPlatform/spanner-migration-tool (HarbourBridge) — core contributor (as pratikdhanave-searce, Searce engagement). Built backend APIs for the Intelligent Schema Assistant, CDC pipelines (Datastream · Pub/Sub · Dataflow), and improved post-migration query performance through PK design, indexing, and table interleaving.
• leopardslab/gocloud — 13 merged PRs to a unified API library for managing cloud resources across AWS, GCP, and Azure.
• New upstream merges (Nov 2026) — first merged fixes into official Apple (coreai-models #503, deterministic preset ordering), AWS (aws-durable-execution-sdk-go #154, O(n) telemetry re-render fix), and two more Microsoft repos (agent-framework #10105, scope docs fixes).
• skycoin/cx — 17 merged PRs enhancing the CX interpreter for deterministic blockchain transaction execution; implemented a Pratt parser reducing parsing complexity by ~30%.

Languages — Go · Python · TypeScript · SQL

Multi-Agent AI — Google ADK · Microsoft Agent Framework · Multi-Agent Reference Architecture · MCP · A2A · LangGraph · LangChain · CrewAI · AutoGen · Pydantic AI · AWS Strands · smolagents · Genkit · Function Calling · ReAct · LLM-as-Judge · Agent Workflows · Cost-aware Constraints

Healthcare AI — FHIR R4 · HL7 v2 · MAI-DxO · SD-Bench · Cost-aware Diagnostic Budgets · Clinical Reasoning

GenAI / ML — Vertex AI Agent Engine · RAG · GraphRAG · LlamaIndex · BigQuery Knowledge Graph · Voice AI (ElevenLabs) · LangSmith · OpenAI · Anthropic Claude · Google Gemini · AWS Bedrock · Hugging Face · Ollama · MLOps · LLMOps

Cloud platforms — GCP (Vertex AI Agent Engine, BigQuery, Cloud Spanner, Cloud Run, Dataflow, Datastream) · AWS (EC2, EKS, Lambda, Bedrock) · Azure (App Service, Cosmos DB, AI Search)

Databases & vector search — PostgreSQL · pgvector · Cloud Spanner · BigQuery · AlloyDB AI · Pinecone · MongoDB · Redis

APIs & protocols — REST · gRPC · GraphQL · FastAPI · Gin · Django · OpenTelemetry · Pub/Sub · Kafka · Apache Airflow

Infrastructure & DevOps — Kubernetes · GKE · EKS · Helm · Docker · Terraform · GitHub Actions · Jenkins · Cloud Run · AWS Lambda

Voice AI — ElevenLabs · Dialog Management · Multi-language Support

IoT & edge — MQTT · OPC-UA · Edge-to-cloud Orchestration

Observability — OpenTelemetry · Prometheus · Grafana · Jaeger · Cloud Trace

Security & compliance — Cloud IAM · KMS · HSM · Secret Manager · Cloud Armor · VPC Service Controls · IAP · Cloud DLP · Binary Authorization · Security Command Center · JWT · OAuth 2.0 · OpenID Connect · Zero-Trust · HIPAA · SOC 2 · ISO 27001 · PCI-DSS · KYC/AML

FinOps — BigQuery cost optimization · Slot-based capacity planning · Query profiling · Materialized views

Domain expertise — FinTech · Open Banking (UAE ADGM/DIFC, Saudi SAMA) · Healthcare Tech · Energy & Utilities · Enterprise AI Adoption · AI Governance & Compliance

Multi-Agent AI · agent memory · enterprise AI adoption · Google ADK · MCP · A2A · Agentic RAG · diffusion & generative media · LLM inference & serving · fine-tuning · vector search internals · distributed systems · database internals · web identity (OAuth/OIDC/SAML) · observability · computer networking · RegTech & compliance · AI cost optimization · on-device AI · the agent-framework landscape (LangChain, LangGraph, CrewAI, Pydantic AI, Strands, smolagents, LlamaIndex) · and the Go, Python, and Rust language curricula — with a focus on the production realities of building reliable AI systems.

• Google Cloud Generative AI Leader — Professional Certificate (Coursera)
• AWS Certified Generative AI Developer — Professional (Udemy)
• Google Cloud Specialization — Data Engineering, Big Data, and Machine Learning on GCP
• Ardan Labs — Ultimate Go and Ultimate Services Go

Looking for senior engineering talent with deep backend + GenAI + multi-region compliance experience? Let's talk.

_{Last updated: November 2026 · Built with care, not with templates.}