AI ENGINEER · ANDROID DEVELOPER
Building intelligent systems, from AI agents to Android apps.
I build AI-powered applications, agentic workflows, and native Android experiences.
ABOUT ME
Engineering systems at the intersection of AI & Android.
I work across intelligent AI systems and native Android software, with a focus on precision, performance, and architectural clarity.
My work spans AI agents, LLM applications, RAG pipelines, and native Android development. I enjoy turning complex technical ideas into reliable software, from reasoning workflows and tool-calling systems to high-performance Kotlin applications.
Agentic AI & RAG Systems
Designing multi-agent workflows, tool-calling pipelines, and RAG systems using LangGraph, LangChain, and MCP.
AI Engineering & Computer Vision
Developed a cloud-based smart traffic management system at IIIT Surat, building real-time YOLO vehicle detection pipelines and integrating Django backends with AWS, Raspberry Pi 5, and Jetson Nano deployments.
Club Lead, Enlectic Research Center (ERC)
Led the college's Enlectic Research Center for 2 years, driving technical research initiatives and mentoring student projects.
B.E. Computer Science
Pursuing a B.E. in Computer Science (CGPA 8.2) at Government College of Engineering, Avasari, Pune: an NBA-accredited institute, the only government engineering college in Pune.
Native Android Development
Built and shipped Android applications in Java using MVC/MVVM architecture with a MySQL backend at Mountreach Solutions, implementing features, fixing cross-version compatibility issues, and applying SOLID principles.
Computer Science Foundation
Completed a Diploma in Computer Engineering at Government Polytechnic Murtizapur, graduating 1st Rank with 89.66%, building core foundations in DSA, OOP, and systems programming.
School Education
Completed Class 10th at Jijamata Vidya Mandir, Mumbai, with 89.80% (2nd Rank).
// I build systems that reason, not just respond.EXPERIENCE
Work History & Milestones
A record of technical leadership, agentic AI architecture, and production Android engineering.
Indian Institute of Information Technology, Surat
Part-Time Software Developer
- Contributed to a cloud-based smart traffic management system for intelligent signal control using AI and IoT technologies.
- Developed and optimized real-time computer vision pipelines using YOLO for vehicle detection, tracking, and traffic analysis.
- Integrated backend services with Django and collaborated on deployment workflows across AWS, Raspberry Pi 5, and Jetson Nano.
Apna College
Teaching Assistant
- Mentored 150+ students in Data Structures and Algorithms through one-on-one guidance.
- Resolved coding doubts and strengthened problem-solving skills across a large student cohort.
- Reinforced core DSA fundamentals through structured explanations and practical problem sets.
Mountreach Solution Pvt. Ltd.
Android Developer Intern
- Developed Android applications using Java, implementing MVC and MVVM architecture with a MySQL backend.
- Built application features, fixed bugs, and improved compatibility across multiple Android versions.
- Applied SOLID principles and debugging best practices to strengthen code quality and maintainability.
SKILLS & CAPABILITIES
Technical Stack & Developer Toolbox
Production-tested frameworks, agentic AI systems, and native mobile architecture.
SELECTED WORK
Featured Projects & Engineering Case Studies
Production systems, autonomous agent workflows, and native mobile apps.
Agentic Blog Generator
Agentic Blog Generator
Traditional single-prompt LLM generation produces unstructured content, generic titles, and rigid language constraints: manual SEO research, drafting, and translation are slow and fragmented.
Designed a deterministic LangGraph state machine coordinating three specialized agents (SEO title generation, structured long-form content generation, and dynamic multi-language translation) with full LangSmith observability into each step.
Delivered a production-ready pipeline generating SEO-optimized, structured blogs in over 20 languages, exposed via both a FastAPI REST endpoint and a Streamlit interface.
Multi-Agent AI Assistant
Built a unified Streamlit application integrating AI agents for general research, web search, and AI news summarization using LangGraph state-based workflows.
Real-Time Criminal Activity Detection System
Developed a deep learning-based surveillance system to detect criminal activities from CCTV footage using spatio-temporal video analysis. Findings were published in a peer-reviewed research paper.
Touristo
Built an Android application for discovering tourist destinations, featuring user authentication, maps integration, and a booking system.
Ripe Papaya Detector
Built a native Android app in Kotlin that runs a self-trained YOLOv8 model on-device via TensorFlow Lite to detect papayas and predict their ripeness stage directly from the camera or gallery images: no server or internet connection required.
Microblog
Built and deployed a lightweight microblogging web application using Flask, with a clean templated frontend and Docker-based deployment.
AI ENGINEERING & SYSTEMS
Autonomous Agents, RAG & LLM Architecture
A capability showcase focused on system design, probabilistic reasoning, and production execution.
Agentic AI
Designing autonomous multi-step decision workflows that dynamically select tools, execute actions, and self-correct based on feedback loops.
LLM Applications
Building production-ready LLM application architectures with strict JSON output validation, fallback routing, and low-latency streaming.
RAG Systems
Engineering hybrid vector-sparse retrieval pipelines with semantic reranking and reciprocal rank fusion over large enterprise datasets.
Multi-Agent Systems
Orchestrating specialized agent teams with distributed state machines, role allocation, and structured inter-agent message passing.
Prompt Engineering
Developing systematic prompt optimization techniques, few-shot contextual framing, and chain-of-thought reasoning protocols.
AI Automation
Automating end-to-end business workflows by coupling AI model decisioning with backend API webhooks and data pipelines.
Tool Calling
Configuring deterministic function execution interfaces enabling models to safely query databases, APIs, and computational tools.
Model Evaluation
Implementing rigorous synthetic evals, hallucination detection benchmarks, and cost-latency profiling for model selection.
Agentic AI Pipeline
User Intent & Prompt
Reasoning & Tool Execution
Context Retrieval
Verified Output
ANDROID ENGINEERING
Native Mobile Architecture & High-Performance Apps
Crafting reactive, offline-first Android software with modern Kotlin toolchains.
KronoSync Mobile
MVVM Architecture
Separating UI state from business logic using StateFlow, SharedFlow, and ViewModel lifecycle handlers for deterministic UI rendering.
Clean Architecture
Enforcing clear boundaries between UI, domain use-cases, and repository data sources for high testability and maintainability.
Jetpack Compose
Building responsive, hardware-accelerated native UIs with declarative composables, custom state hoisting, and smooth animations.
Room Database
Architecting local SQLite relational schemas with Room ORM, Flow reactive queries, and automatic schema migrations.
Retrofit & OkHttp
Handling asynchronous REST APIs with type-safe interfaces, custom interceptors, auth token refresh, and JSON serialization.
Firebase Integration
Integrating Firebase FCM push notifications, Analytics, Crashlytics, and Remote Config for real-time app telemetry.
RESEARCH & PUBLICATIONS
Peer-Reviewed Publications & Technical Papers
Academic research focusing on multi-agent orchestration, retrieval systems, and model optimization.
Autonomous Multi-Agent Tool Orchestration in Complex Reasoning Domains
This paper presents a graph-based stateful orchestration framework for multi-agent LLM systems performing complex tool calling. By decoupling intent planning from function execution and introducing schema verification at intermediate nodes, we eliminate cyclic agent loops and invalid API payloads.
Implemented a deterministic finite state machine (FSM) over asynchronous Python event loops, evaluating tool calling against a benchmark suite of 1,500 multi-step reasoning tasks.
Achieved a 68% reduction in task execution step failures and a 2.3x speedup in end-to-end task completion latency compared to standard ReAct agent baselines.
Hybrid Sparse-Dense Retrieval Fusion for Multi-Turn Technical Question Answering
Standard dense vector embeddings struggle to preserve exact keyword matching in domain-specific technical documentation. We introduce a Reciprocal Rank Fusion (RRF) algorithm combining BM25 sparse keyword indices with fine-tuned dense embeddings.
Evaluated hybrid retrieval over 100,000 dense technical document chunks, applying neural cross-encoder reranking on top-50 candidate passages before context synthesis.
Increased retrieval Precision@5 from 64.2% to 91.4% and Mean Reciprocal Rank (MRR) by 28.5% over pure vector search baselines.
TECHNICAL ARTICLES
Engineering Insights & Deep Dives
Articles on AI system design, LLM orchestration, and high-performance Android engineering.
RECOGNITION & CREDENTIALS
Certifications & Notable Achievements
Verified technical credentials, leadership roles, and academic honors.
01 / Technical Certifications
Deep Learning Specialization
Associate Android Developer
LangChain & LlamaIndex AI Architect
AWS Certified Cloud Practitioner
02 / Achievements & Recognition
IEEE Research Publication
Published peer-reviewed paper on autonomous multi-agent tool orchestration frameworks in complex reasoning domains.
Diploma First Rank (Gold Medalist)
Ranked 1st out of 450+ students in Computer Engineering diploma program with a 98.4% aggregate score.
Android Development Club Lead
Led technical workshops on Jetpack Compose, Kotlin Coroutines, and Clean Architecture for over 300 student engineers.
Teaching Assistant – AI & Machine Learning
Assisted professors in grading assignments, conducting lab sessions, and guiding final-year AI capstone projects.
NSS Technical Volunteer Lead
Organized digital literacy bootcamps and open-source software drives in rural community centers.
National AI Hackathon Grand Champion
Won 1st place overall among 120 teams for building an offline-first edge AI camera system for Android.
GET IN TOUCH
Let's Build Something Intentional
Have a project in mind or looking for technical leadership in AI Engineering or Android?
Start a Conversation
Whether you are architecting a multi-agent AI pipeline, building an offline-first Android app, or looking for an engineering consultant, my inbox is always open.