Hi, i'm Aayush
Hi, i'm Aayush
AI Engineer building production-ready AI systems, agents & intelligent products.
Agentic workflows, vision pipelines, developer tools & full-stack systems.
About me
AIEngineerspecializinginagenticsystems,computer-visionpipelines,developertools,andfull-stackAIproducts.Witharigorousengineeringfoundationspanninghardwarefundamentals,machinelearning,andmodernwebarchitectures,Iengineerend-to-endresilientsystems—fromlocaledgemodelstoproductionmulti-agentswarms.
B.Tech, Electronics and Computer Engineering
Engineering Proof & Systems Verification
LangChain, Qdrant Vector DB, BM25 Hybrid Search, Gemini API, Ollama Local Models, PyTorch
YOLOv8, BoT-SORT, OpenCV, TensorFlow Lite INT8, Qwen2.5-VL, Grad-CAM Saliency Maps
FastAPI Async Workers, Python, PostgreSQL, REST Architecture, Atomic File Persistence
Next.js 14/15, TypeScript, React, Semgrep AST Security, VS Code Extension APIs, Docker
Journey
A hybrid engineer with a software-first direction: hardware fundamentals → AI systems.
Electronics & Embedded Hardware
Hardware fundamentals: Microcontroller architecture, RTOS, sensor interfacing, and C/C++ data structures. Built low-level firmware foundations.
Python, ML & Computer Vision Pipelines
Machine learning & deep vision: Deployed real-time YOLOv8 vehicle tracking, INT8 quantized edge models, and predictive ML classification pipelines.
AI Engineering, RAG & Full-Stack Systems
Production systems: FastAPI microservices, vector databases (Qdrant), offline-first PWAs, and industrial Hardware-in-the-Loop (HIL) validation at SEDEMAC.
Agentic AI, IR & Control Plane Systems
Autonomous systems: Multi-agent triage swarms, deterministic agent control planes (SENTINEL-X), hybrid ranking platforms (SearchOps), and local VLMs.
Services
AI & Agentic Systems
Multi-agent autonomous swarms, hybrid retrieval RAG pipelines, Qdrant/BM25 integration, and local or cloud LLM orchestration tailored for enterprise workflows.
Computer Vision & Edge AI
Real-time vehicle and object tracking with YOLO/OpenCV, INT8 quantized TFLite models for edge devices, and offline bilingual Vision-Language models.
Full-Stack AI Products & APIs
Production Next.js, React, TypeScript, FastAPI, and Node.js web applications built with ruthless speed, fluid responsive animations, and rock-solid database schemas.
Developer Tools & Automation
Custom CLI utilities, VS Code extensions, automated static security analysis gates, and continuous deployment workflows that multiply engineering velocity.
Hardware & Embedded Foundation
Firmware engineering for ESP8266, Raspberry Pi, and Arduino, combined with hardware-in-the-loop (HIL) testing and IoT telemetry dashboards.
Projects
Engineered AI systems spanning multi-agent autonomous swarms, offline edge models, developer security tooling, and full-stack platforms.
SRE & FinOps Triage Swarm
Multi-agent autonomous incident investigation and cloud-budget-aware troubleshooting using hybrid retrieval with Qdrant + BM25.
65% reduction in simulated incident triage latency compared to manual correlation across 3+ dashboards.
SENTINEL-X
AI Agent Reliability, Security & Verification Control Plane — local-first runtime enforcing deterministic tool policies, secret redaction, and CI/CD quality gates.
44/44 deterministic passing tests covering policy enforcement, red team attack simulations, and secret interception.
SearchOps
Information Retrieval, Hybrid Ranking & Search Engineering Platform with real-time score decomposition and automated IR benchmark suite.
0.922 Recall@10 and 1.000 MRR benchmarked on local sentence-transformers (all-MiniLM-L6-v2) + Cross-Encoder reranking.
Recon AI
Disaster-response system designed around offline-first communication and store-and-forward networking during connectivity failures.
100% offline data persistence across device reboots and simulated complete network outages.
Offline-first VLM pipeline for extracting structured JSON from complex bilingual Marathi/English documents on local CPU infrastructure.
100% private on-premise execution with zero outbound internet network calls.
Skill Map
Verified competencies backed by concrete production repositories and deployed architectures.
AI & Agentic Systems
Experience
Industrial hardware-in-the-loop validation, machine learning research, and full-stack software delivery.
Independent AI Engineer & Software Builder
- ·Architecting custom AI agentic workflows, multi-agent swarms, and hybrid vector RAG systems.
- ·Building full-stack web products, robust developer tooling, and automated LLM pipelines.
- ·Consulting on computer vision and edge AI deployment for local/offline constraints.
Product Support Engineer Intern
- ·Hardware-in-the-loop (HIL) validation for automotive ECU controllers.
- ·Diagnostics, fault code analysis, and sensor anomaly tracking across mechatronic systems.
- ·Automated quality testing workflows and performance metric logging.
Machine Learning Intern
- ·Python machine learning pipeline development, data preprocessing, and feature engineering.
- ·Model training, hyperparameter optimization, and predictive evaluation metrics.
- ·Delivered clean analytical outputs aligned with production performance constraints.
Student Developer Intern
- ·Low-level firmware instructions on microcontrollers for industrial automation.
- ·Interfaced peripheral sensor matrices and communication protocols.
Event Head
- ·Organized university-wide technical bootcamps, hackathons, and hands-on workshops.
- ·Trained and mentored over 1,500+ engineering students in modern development and AI.