Research & Engineering

AI agent infrastructure, from prototype to production.

What I'm building

LLM agents need infrastructure that is efficient (long contexts, memory optimization, redundancy reduction) and secure (least-privilege access, adversarial resistance). I build at this intersection: context optimization, authorization, memory management, and observability.

Currently exploring

  • MCP Security - OWASP MCP Top 10 practitioner coverage, shadow MCP detection, fine-grained authz
  • Agent Identity - Non-human identity lifecycle for AI agents (NHI problem)
  • Context Engineering - Distill 1.0 with MCP integration; ThinkBudget adaptive inference budgets
  • agent-trace v0.4.0 + v0.5.0 - production CI mode, OTLP export, container instrumentation
  • Speaking - AI Engineer World's Fair CFP submitted; KubeCon NA 2026 on the radar

Focus areas

1

Context Efficiency & Reliability

"How do we make LLM outputs reliable and deterministic through better context management?"

Clean, deduplicate, and optimize context before it reaches the model. Deterministic algorithms over probabilistic heuristics.

DistillContextLabTokenVMKV-Cache Profiler
2

Agent Authorization & Audit Trails

"How do we enforce fine-grained, capability-based authorization for AI agents with full auditability?"

Google Zanzibar-style authorization for agent-tool interactions via OpenFGA and MCP. Dynamic capability tokens, real-time policy enforcement, audit logging.

agentic-authzAgentic AuthorizationA2ASOpenFGA
3

Adversarial Robustness & Observability

"How can agents maintain safety under prompt injection, tool poisoning, and adversarial tool responses?"

Observability and tracing to detect attacks on agent tool-use pipelines. GPU-level profiling for performance and security.

Sentinel AIactionsecLLMTraceFXAgentflow

Products & Prototypes

Context Engineering

Distill

Deterministic context deduplication for LLMs in ~12ms

Agent Authorization

agentic-authz

OpenFGA + MCP authorization gateway for AI agents

Observability

LLMTraceFX

GPU-level LLM inference profiler with AI-powered insights

Authorization

OpenFGA

Google Zanzibar-style authorization (Core Maintainer)

Context Engineering

ContextLab

LLM context engineering toolkit - analyze, compress, visualize

Memory Systems

TokenVM

KV-cache virtual memory with page-based eviction

Agent Systems

Agentflow

Agent orchestration with capability-based security

Security

Sentinel AI

Security scanning with LLM-powered triage

Security

A2AS

Agent-to-agent security protocol implementation

DevSecOps

actionsec

GitHub Actions security analyzer

Working with me

Systems engineering experience meets research curiosity.

  • Ships: research prototype to production. Built Distill, maintain OpenFGA
  • Systems depth: production infrastructure at Ona (formerly Gitpod)
  • Open source: OpenFGA maintainer (CNCF), GitHub1s maintainer
  • Cross-domain: security, distributed systems, ML infrastructure
  • 40+ articles on siddhantkhare.com/writing

Let's talk

Book a call Resume Research statement
Siddhant Khare
Agentic Engineering Guide Resume