Skills
Grouped by the layer of an agent system I work on, with the work each one rests on.
Tools and protocols
- Designed and run two MCP servers (pi-mcp and the-wardrobe): small, well-named tools with checks before writes.
- Led a type-information API that serves Apex code metadata (class, method and field signatures, generics, API version bounds) to AI code-generation and developer tools.
- OAuth per client and per-path access rules for agent memory: what can be written freely, what needs approval, what is locked.
Context and grounding
- Onboarded Apex metadata into a metadata catalog and the grounding service, so LLM-powered code features get accurate context through MCP.
- Moved AI code-generation and test-generation skills onto the new API and removed their dependency on an external MCP server.
- Versioned memory with conflict checks, so two agents cannot silently overwrite each other, and human approval for edits that matter.
Reliability and operations
- Profiled and diagnosed a tail-latency problem in compilation workloads with JFR and CPU profiling; added lock-contention telemetry, SLO dashboards and regression alerts.
- Distributed-cache anomaly alerts, on-call rotations, root-cause playbooks, feature-gate cleanups and release backports.
- At Sprinklr: moved 500+ microservices from VMs to Kubernetes, built a MongoDB metrics exporter (Prometheus, Kafka, Grafana), built resource-utilization tooling that cut cloud cost by about 20%, and automated disaster recovery across 2,000+ database servers with a team of ten.
- This site's own assistant is a small example: rate limits, a monthly spend cap, an instant off switch, provider fallback and logs with personal data redacted.
Agent features
- Refactored AI code-generation and test-generation skills to read from the metadata API directly.
- Multi-agent research system: specialised agents that report to a coordinating layer (described at a high level only).
- Defences in a public assistant: untrusted input treated as data, single-turn answers, length caps and an answer scope limited to this site.
Working with others
- Technical subject-matter expert: scoped multi-sprint epics with product and scrum, and wrote design proposals for leadership.
- Mentored junior engineers through pair programming and architecture reviews.
- Ran team-wide demos and workshops that drove adoption of AI-assisted engineering.
Stack
- Java
- Python
- FastAPI
- Spring Boot
- PostgreSQL
- MongoDB
- Elasticsearch
- Kafka
- React
- MCP
- Docker
- Kubernetes
- Prometheus
- Grafana
What this site does not show
- Customer-facing or forward-deployed delivery. My work has been with internal product, scrum and partner engineering teams.
- Model training or fine-tuning.
- Public open-source code: the project repositories are private for now.