Agent Policy Layer
Python · protocol design · CI/CodeQL
An open-source guardrails protocol that enforces agent policies at the call site — runtime-agnostic, fail-closed by default, one line to instrument.
AI engineeragents · evals · guardrails · RAG
Dublin · IE 🇮🇪
Portfolio
Sentinel
Python · SQLite · Open Banking · systemd
A self-hosted personal finance system running continuously on a Linux VPS — it ingests live bank transactions (PSD2), categorizes them deterministically, tracks recurring bills, and pushes a daily safe-to-spend number over Telegram.
Career
Jan 2025 — now · Dublin, IE 🇮🇪
IBM
AI Engineer/Software Developer
Jun — Aug 2024 · Dublin, IE 🇮🇪
Dublin Institute for Advanced Studies
Computational Astrophysics Intern, GPU/HPC
Sep — Nov 2022 · Bangalore, IN 🇮🇳
AI Health Highway · IISc. Bangalore
DSP (bio-signals)/ML Engineer
2018 — 2022 · Chennai, IN 🇮🇳
SRM University
B.Tech Computer Science — Big Data Analytics
Selected study
On personal time, on top of the degrees. The CS224N assignments produced BERT and a minimal GPT-2, from scratch.
Stack
Production — agent systems
watsonx OrchestrateMCPLangGraphLangChainLangSmithLlamaIndexFastMCPLangfuseUnitxtOpenSearch
Production — model providers
OpenAI APIAnthropic ClaudeAWS BedrockOllamaGroq
Methods
Agent evaluationLLM-as-a-JudgeTool use / Function callingStructured outputsEmbeddings / Vector searchRed-teaming / prompt-injection testing
Production — platform & delivery
PythonSQLBashFastAPIgRPCCeleryRedisPostgreSQLDockerKubernetes / OpenShiftTektonGitHub ActionsCodeQLPrometheus / GrafanaOpenTelemetryIBM Cloud
Open source & independent
PyTorchHugging FaceTypeScriptSQLitePSD2 / Open BankingTelegramLinux / systemdPineconeAWS — SageMaker · Lambda · S3 · EC2
Earlier work
C++CUDATensorFlowSparkDSP · Librosa
About
0:00 / 0:00