Building Systems that Scale, Perform, and Deliver

Full-stack developer and technical lead who drives execution from vision to business impact — specializing in scalable, distributed architecture, cloud-native solutions, and ML/AI integration

Technical Showcase & Case Studies

Architectural patterns, system designs, case studies and real-world implementations — each focused on demonstrating specific technical decisions and approaches

Real-Time Conversational AI Agent with Streaming Architecture

Full-Stack AI System Design

Implemented as an in-memory, event-driven, async pipeline, the system uses asyncio queues for inter-stage data flow, configurable event routing, and generator-compatible stages for streaming. Each conversation runs an isolated pipeline with its own async tasks, WebSocket connection, and pluggable backends for STT, LLM, and TTS. A separate control plane provides pub/sub state updates and awaitable control signals for cross-stage coordination. The system supports multiplexing multiple STT/TTS processing over a single backend.

Architecture Patterns

Real-Time StreamingEvent-DrivenAsync PipelineWebSocketPub/SubTool-Use

Technologies

FastAPIAsyncioWebSocketReactSTT/TTS StreamingLLM StreamingLLM Tool Use
Publishing November 17, 2025

SVO-Constrained QA Synthesis with Hybrid Dual-Scale Retrieval

Knowledge Indexing and Technical QA System Design

A pipeline for transforming unstructured technical data into a query-optimized knowledge base. System performs semantic chunking, SVO triplet extraction, and SVO-constrained generation to create precise, low-cost QA pairs. A distilled extractive QA model validates and filters generated pairs, while redundancy is reduced via Affinity Propagation clustering. Retrieval combines dual-scale embedding (QA + chunk level) with hybrid dense, sparse, and late-interaction (ColBERT) scoring for both granular and broad technical search.

Architecture Patterns

SVO ExtractionConstrained GenerationClusteringData PipelineRAGHybrid Search

Technologies

SpacyLLMEmbedding ModelsCross-EncodersQdrantAffinity PropagationIDF

High-Performance Ticket Reservation System

Architecture Design Case Study

This case study focuses on a ticketing system for ultra-large-scale events, handling 100K seats, 1M+ RPS and >10M potential concurrent users (queue). Addresses architectural decisions from multiple angles (DB limits, CAP, UX, Fairness, Exploitation) and discusses tradeoffs, queue management strategies, database partitioning, caching and failure scenarios. Includes load calculations, cost modeling, and scalability analysis across multiple architectural approaches.

Architecture Patterns

Database ArchitectureQueueHigh AvailabilityCQRSEdge computingMulti-stage cache

Technologies

AWSDynamoDBCloudfront FunctionsElasticache / Redis

Serverless Link Shortener with Live Analytics

Distributed Microservices

High-availability microservice based URL shortener with QR generation and real-time analytics dashboard. Built on serverless edge architecture demonstrating CAP theorem tradeoffs, global distribution, and cost optimization. In addition, the project contains an Admin SPA (React).

Architecture Patterns

Event-DrivenDistributed ArchitectureServerlessEdge ComputingHigh AvailabilityMicroservices

Technologies

Cloudflare WorkersDurable ObjectsWebSocketPrisma ORMReact

Fine-Tuning Coder Model with LoRA

ML Model Training

Parameter-efficient fine-tuning of code generation model for domain-specific tasks. Demonstrates understanding of modern ML training pipelines, transfer learning strategies, and Hugging Face ecosystem integration.

Architecture Patterns

MLOpsTransfer LearningModel Training Pipeline

Technologies

Hugging FacePyTorchTransformersLoRA

Analytics & ML Case Studies

Data Science Workflows

End-to-end data science workflows considering real world business objectives in Loan approval, Bike sharing and Telecom industry. Preprocessing pipelines, exploratory analysis, feature engineering with RFE/RFECV, and production ML model training (Random Forest, Gradient Boosting) with hyperparameter optimization.

Architecture Patterns

ML PipelineFeature EngineeringModel Selection

Technologies

PythonScikit-learnPandasStatistical ModelingHyperparameter Tuning

Skills & Technologies

Backend & APIs

Node.jsPythonFastAPINext.js API / RSCRESTful APIsGraphQLWebSocket

Frontend

ReactNext.jsTypeScriptTailwind CSSshadcnMaterial UI

Architecture & Design

Distributed SystemsMicroservicesEvent-Driven ArchitectureServerlessHigh AvailabilityCQRSCAPCDCQueuePub/SubOCCSystem DesignScalability

AI/ML & Data

LLMRAG / GraphRAGHybrid SearchPyTorchHuggingfaceJupyterPandasScikit-learn

Databases

PostgreSQLMongoDBDynamoDBPrisma ORMSQLAlchemyQdrant

Technical Leadership

Team LeadershipProject Scoping & PlanningStakeholder CommunicationCode Review & MentoringRisk ManagementRequirement AnalysisTechnical StrategyCompliance

Cloud

AWS (multiple services)Cloudflare workers

Integration

Third-party APIsOAuth/AuthOIDCWebhooks

About

My approach

My approach centers on establishing lightweight processes that create clarity. I apply agile methodologies through principles, not rigid ceremonies — standups that unblock work, retros that drive change, and planning that serves delivery. I coach teams through complex technical decisions and maintain the engineering standards that turn code into reliable products.

Whether designing distributed systems, integrating ML/AI solutions, or scaling high-availability infrastructure, I focus on delivering features that drive business value and choosing architecture that fits the product stage — from PoC to MVP to scaling for millions of users.

When authority is needed to protect quality or timelines, I use it — but I'd rather influence through collaboration . I can work across product, design, marketing, and operations to align priorities and translate between technical execution and business strategy. I've learned that the best solutions emerge when engineers understand the "why" behind the work, and stakeholders understand the trade-offs in the "how."

Currently pursuing

AWS Solutions Architect Professional and ML/AI Specialization certificates to formalize my ongoing learning and personal development.


Ready to build?

Let's discuss your project

TXTTW