I like building systems where AI meets real engineering constraints β memory, latency, security, distributed data, cloud infrastructure, and cost.
My recent work spans persistent-memory agents, codebase intelligence graphs, multi-agent developer tooling on AMD GPUs, AI-powered endpoint detection and response, and quantitative trading systems.
AI Systems β Agents β’ RAG β’ Memory β’ Model Routing
Backend β FastAPI β’ Python β’ REST APIs β’ Async Systems
Cloud & DevOps β AWS β’ Alibaba Cloud β’ Docker β’ Kubernetes
Security β EDR β’ Threat Detection β’ PII β’ Prompt Injection
Data β DuckDB β’ SQLite β’ PostgreSQL β’ CockroachDB β’ Vector Search
Performance β ROCm β’ HIP β’ Quantization β’ Token Optimization
|
Persistent memory & intelligent routing for LLM gateways Built a memory-augmented LLM gateway with semantic memory, Ebbinghaus-style forgetting curves, semantic caching, and intelligent routing across six Qwen models. Highlights
Hackathon β’ Repository |
Private multi-agent engineering team on AMD GPUs Built a conversational codebase analysis system powered by AMD ROCm/HIP. Six specialized agents collaborate to inspect security, performance, architecture, and DevOps concerns. Highlights
|
|
AI security analyst with multi-layered cognitive memory Built an AI-powered Endpoint Detection and Response (EDR) system with cognitive memory, combining live threat telemetry with historical incident patterns to accelerate investigation. Highlights
Hackathon β’ Repository |
Token-budget-aware agentic inference system Built a containerized general-purpose agent engineered to maximize benchmark accuracy while minimizing total token consumption and inference latency. Highlights
Hackathon β’ Repository |
|
Quantitative options trading daemon with HAR volatility modeling Built an automated quantitative options trading daemon on Alpaca executing defined-risk Iron Condors on SPY & QQQ, driven by an Enhanced HAR volatility forecast model and strict capital kill-switches. Highlights
|
Deterministic change-impact analysis & codebase knowledge graph Built an incremental repository intelligence system that maps codebase dependencies into a local DuckDB graph, delivering sub-millisecond impact traversal at $0 API cost alongside cited AI explanations. Highlights
Repository β’ Demo |
Arrays / Strings ββββββββββββββββββββ
Hash Tables βββββββββββββββββ
Dynamic Programming βββββββββββ
Math βββββββββββ
Sorting βββββββ
Backtracking ββββ
Divide & Conquer βββ
Database / SQL βββββββ
LeetCode profile: aamira6205 Β· 423 C++ problems Β· 65 MySQL problems Β· 10 JavaScript problems
LangChain LangGraph RAG Vector Search Embeddings LLM Agents
Model Routing Semantic Caching Prompt Engineering
Alibaba Cloud AWS ROCm HIP CI/CD REST APIs
|
Agentic AI Memory, reasoning, routing, RAG and multi-agent orchestration. |
Cloud Containerized services, Kubernetes, AWS and Alibaba Cloud deployments. |
Security EDR, threat detection, PII protection and prompt-injection defense. |
Performance GPU inference, quantization, caching and token-efficient systems. |
Building production-grade AI infrastructure, especially systems where model intelligence is only one part of the problem.
I'm particularly interested in:
Distributed AI Infrastructure
β
Agentic Systems + Persistent Memory
β
Security + Reliability
β
Cloud-Native Deployment
β
FinTech / High-Impact Applications
