Dhruv Mulay
AI Research EngineerAI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI Research Engineer with less than a year in AI/ML systems and Full-stack Software Engineering
B.Tech Computer Engineering graduate from Vishwakarma Institute of Technology, Pune (graduated June 2026, final CGPA 8.11 / 10), currently on a research exchange semester at Hochschule Aalen, Germany. Specialisation in AI/ML systems and full-stack software engineering: production computer-vision pipelines, video diffusion models, agentic LLM architectures, and applied information-theoretic / statistical research. Three internships spanning academic research and production AI. Hackathon winner and active competitive-ML practitioner (Kaggle, Zindi, Yandex Cup, CISPA).
Hochschule Aalen (University of Applied Sciences)
Research Exchange Semester
August 1, 2025 – Present
Vishwakarma Institute of Technology (VIT)
B.Tech · Computer Engineering
August 1, 2022 – June 30, 2026
Ryan International School
ICSE (Class X)
N/A – May 31, 2020
Kaveri College of Arts, Science and Commerce
Higher Secondary Certificate (HSC) · Science - PCM
N/A – May 31, 2022
Genzoic
Fullstack Development & AI Intern
July 1, 2025 – December 31, 2025
Pune, Maharashtra, India
Hochschule Aalen (University of Applied Sciences)
Research Intern - Geometric / Information-Theoretic Neural Feature Learning
March 1, 2025 – Present
Aalen, Baden-Württemberg, Germany
Rikaian Technology Pvt. Ltd.
R&D Intern
August 1, 2024 – November 30, 2024
Pune, Maharashtra, India
Autonomous Driving Scene Generation - Yandex Cup ML
June 24, 2026 – Present
Controllable video scene generation for autonomous driving using the CogVideoX-2b diffusion model. Custom STDIT3-XL/2 architecture with camera-angle embedding and bounding-box conditioning (cars, trucks, buses, people, traffic signs). Flash Attention + xformers optimisation; Docker + DVC reproducible submission pipeline.
View ProjectMarketPulseTracker - AI Stock Sentiment Platform
June 24, 2026 – Present
Real-time stock sentiment with Google Gemini AI; 5-day momentum scoring and AI pulse ratings (bullish/bearish). In-memory caching (10-min TTL); Kubernetes deployment; GitHub Actions CI/CD.
View ProjectAutomated Financial Data Analysis & Interactive Dashboard
June 24, 2026 – Present
ETL pipeline cleaning 5+ years of stock data across 50+ companies (P/E, ROE, EPS standardisation). Calculators for 10+ financial metrics (volatility, Sharpe ratio), cutting manual error by 85%. Dash dashboard with filters, charts, heatmaps; auto-refreshed daily via cron (analysis time ~70%).
Edge-Optimised ML Pipeline - Pune Metro Surveillance (2nd Place, Maha Metro Hackathon)
June 24, 2026 – Present
Stacked YOLO-SAM pipeline for real-time safety monitoring (50K+ images, VGG16, 50+ epochs). Parallel stream processing: 8 simultaneous 720×480@30fps feeds, 240fps aggregate throughput (<60ms latency, 79% accuracy, 15% reduction in false positives).
Interpretable Credit-Risk Model - Retail Loan NPA / Default Forecasting
June 24, 2026 – Present
Trained Logistic Regression (85% accuracy, 0.87 AUC-ROC) and SVM (89% accuracy, 0.89 AUC-ROC) on an imbalanced default dataset (5-10% positive). Cut feature set by 40% with <2% AUC drop; validated calibration with Brier score and ±3% reliability plots; identified debt-to-income ratio as dominant predictor (r=0.63).
Hey Jarvis - On-Device Voice-to-Code Assistant
June 24, 2026 – Present
On-device speech pipeline: wake-word ("Hey Jarvis") → VAD-gated transcription → INT8-quantised Whisper (whisper.cpp / ggml) → LLM tool trigger. Latency-tuned for sub-second wake-to-text on consumer laptop hardware (no GPU).
Scalable Real-Time Chat Application
June 24, 2026 – Present
WebSocket backend: <100ms latency, 5K+ concurrent users, STOMP protocol; JPA/Hibernate + MySQL persistence. RabbitMQ pub/sub: 99.9% delivery reliability, 10K+ messages/sec; RESTful APIs with JWT auth (<50ms).
LLM-OS - Multi-Agent AI Operating System
June 24, 2026 – Present
Orchestrates 14 specialised agents (Gmail, Calendar, browser automation, code generation, ML experiment management) via a task-DAG planner with circuit breakers and event streaming. FastAPI + Celery + Redis backend; ChromaDB RAG semantic memory; WebSocket event streaming. Hybrid LLM: local Phi-4-mini (llama.cpp) + Gemini 2.5 Pro; "LLM Council" multi-model deliberation. Tauri 2 desktop GUI: React 19 + Three.js 3D visualiser; Telegram bot interface.
View ProjectCISPA Hackathon 2026 - AI Privacy & Security Research
June 24, 2026 – Present
Audited 5 state-of-the-art AI-security papers (ICML 2023, USENIX Security 2023, NeurIPS 2024, ICLR 2025) for reproducibility and correctness. Identified and patched a missing Šidák p-value combination (Eq. 2) in the official NeurIPS 2024 dataset-inference repository. Built a black-box adversarial extraction pipeline against Stable Diffusion v1-4; reproduced a memorisation threshold (intra-clique L2: 0.0213 vs <0.15).
View ProjectLinkedIn Job Hunter Automation
June 24, 2026 – Present
AI-powered LinkedIn job scraper and personalised outreach tool; browser extension + BeautifulSoup pipeline.
View ProjectIBM Full Stack Development Certificate
IBM
June 1, 2026 – Present
NPTEL Mathematics for Engineers
IIT Madras
June 1, 2026 – Present
Cultural Fit Analysis
The candidate exhibits a strong cultural fit for a research-oriented and innovative environment. Their diverse projects span autonomous driving, financial analysis, real-time chat, and multi-agent AI systems, showcasing adaptability and a broad interest in applying AI to various domains. Participation in hackathons and competitive ML platforms (Kaggle, Zindi, Yandex Cup, CISPA) indicates a collaborative and competitive spirit, as well as a willingness to tackle challenging problems. The research exchange semester further demonstrates an openness to international collaboration and diverse academic settings. The focus on reproducible pipelines (Docker, DVC) and CI/CD suggests an appreciation for robust engineering practices.
Soft Skills & Operational Fit
The candidate demonstrates strong initiative and leadership through extracurricular activities (Cultural Head, Viculp) and participation in competitive hackathons. Their ability to manage teams and reduce event setup time by 40% indicates good organizational and efficiency skills. The detailed project descriptions suggest a methodical approach to problem-solving and a focus on measurable outcomes. The research exchange semester and competitive ML participation highlight a proactive learning attitude and a drive for continuous improvement, which are valuable for an AI Research Engineer role.