Meet your next
creative AI engineer.

I build the AI platform
your business runs on

for engineers
and everyone else.

I love building
things that scale

and I ship it
every time.

Trading
Research
Geopolitics
Fitness
Animals
Temples
Music
Trekking
Songs
Strays
Lifting

and still manage
to make free time.

I’m up for new

adventures

and I’d like to
work with you.

Contact me

Recent work and Research

OakNorth

2026 – Present

Senior Technical Lead

I build OakNorth's internal AI platform — a self-serve gateway that lets engineers and non-engineers alike provision and ship AI. I built the skill and MCP registries, deployed agents on Amazon Bedrock AgentCore with models served through AWS Bedrock, and shipped the platform end to end: a Next.js UI, a FastAPI backend, and role-based access control (RBAC). I use Terraform extensively to provision everything running behind it — agents, Git repositories, apps, containers, and ECS tasks — across AWS, with Aurora as the data layer.

Programming Languages

PythonJavaTypeScriptPySparkPyTorchCUDA

AI/ML Tools & Frameworks

Stable DiffusionComfyUIInvokeAIFastMCPMCP ServerLiteLLMPineconeLangChainLangGraphRAG PipelinesDBSCANK-MeansModel TrainingInference Services

Model Fine-Tuning

UnslothLLaMA-FactoryDeepSpeedAxolotlTRLPEFTLoRAQLoRADPOGRPOSFT

AI Agents

Amazon Bedrock AgentCoreMCPA2A Protocoln8n AgentVS Code Agent

Full-Stack Frameworks

SpringSpring BootAngularNext.jsFastAPI

ETL & Data Tools

AirflowDatabricks WorkflowsTrinoPrometheusLokiGrafanaGCP BigQuery

Packages & Libraries

PandasNumPyScikit-LearnBoto3SeabornJFrog

Databases

MySQLPostgreSQLRedisElasticsearchPinecone

Messaging & Queues

KafkaSQS/SNS

DevOps & CI/CD

Kubernetes (K8s)DockerTerraformGitLab CI/CDGitLab ActionsKubecost

Cloud & AWS Services

S3EC2RedshiftRDSAPI GatewayCloudWatchLambdaEBSEKS

Previous experience

Electronic Arts

2022 – 2026

Software Engineer 2 – AI Platform

I designed and developed an AI platform where engineers, designers, and data scientists provisioned text, image, and voice models alongside integrations like diffusion models, MCP servers, agent-to-agent communication (A2A protocol), and vector databases. I owned the full lifecycle: system design → UI → backend → ML model training & inference services → custom CUDA kernels → Kubernetes deployment & GPU optimization.

Demandbase

2020 – 2022

Software Engineer 1 – Backend

Built microservices from scratch with Spring Boot, migrated Elasticsearch clusters, developed REST APIs, and managed AWS deployments with CloudWatch observability — the backend foundation that lets me own the full stack today as an AI platform engineer.

Applied Information Sciences

2019

Software Engineer

Full-stack development — Angular on the frontend, C# on the backend, and SQL for data.

Research

Published papers on AI, cloud computing, and optimization — Google Scholar

CAMQU

NP-Hard Cloudlet Allocation via Multilevel Queues

Solved the NP-Hard Cloudlet Allocation problem by reducing time complexity to O(log n) using multilevel queue scheduling. Outperformed FCFS by 38.2% and SJF by 25.6%.

Read paper →

Hybrid Recommendation System

ML-Powered Recommendations on Hadoop MapReduce

Built a recommendation engine combining collaborative and content-based filtering with Alternating Least Squares on Hadoop MapReduce — improving accuracy by 30%.

Read paper →

Live Migration Annealing

AI-Powered VM Live Migration via Simulated Annealing

Developed a simulated annealing algorithm with real-time time series analysis for VM live migration — cutting required iterations by 50% and migration time by 70%.

Read paper →

SLA-based Cloudlet Allocation

ML-Driven Allocation Using ACO & A* Search

Solved NP-Hard cloudlet allocation in heterogeneous environments using Ant Colony Optimization, A* search, and TSP heuristics — reducing time consumption by 47%.

Read paper →

Energy Aware VM Migration

K-Means Driven VM Consolidation for Datacenters

Decreased datacenter energy consumption by 28% through K-Means clustering on power metrics for intelligent VM consolidation and migration.

Read paper →