Building smart traffic GNNs, ML security pipelines, and automated cloud workflows.
Ask Dax anything about Daksh's projects, experience, or patent to get verified information.
Production-grade builds, data science models, and cloud-native environments.
Designed an end-to-end smart traffic perception system with YOLOv8-based vehicle and emergency vehicle detection, real-time lane density estimation, and topological road network modelling as a graph using NetworkX.
Scalable multi-layer intelligence · YOLOv8 Emergency Priority · Graph Network Modeling
NETRA uses a three-tier model: a Perception Layer using YOLOv8 to classify normal vs. emergency vehicles and calculate lane occupancies; a Graph Layer using NetworkX to map traffic intersections as nodes and lanes as edges; and a Control Layer executing emergency prioritization logic to dynamically optimize signal timings.
Graph structures were selected instead of standard grids to facilitate coordinate-independent road network representations. This enables scale-invariant routing, allowing the control system to easily adapt from small localized intersections to large-scale city grids.
Real-time execution suffered from high GPU frame processing times during heavy vehicle counts. This was mitigated by framing downsampling, executing vehicle detection selectively every 3 frames while relying on light optical flow tracking for intermediate frames.
Designed a full-stack workflow intelligence platform with React 19, TypeScript, and Express 5. Built a visual rule engine using json-rules-engine with client-side OCR (Tesseract.js WebAssembly), webcam KYC photo capture, drawn signatures, and version layout snapshots.
Visual Logic Canvas · Client-Side OCR (WASM) · Auto Lead Routing & Sentiment
Designed a multi-tier platform: a React 19 + TypeScript + Vite client dashboard containing a drag-and-drop workspace and rule graph canvas; an Express 5 REST API executing automatic scoring, tag classification, and sales rep assignment; and an SQLite 3 storage layer mapping schemas, audit logs, and version control.
Chose Tesseract.js (running locally via WebAssembly) to parse uploaded user IDs and document proofs on the client side. This eliminates external server transmission of sensitive user data, matching KYC compliance constraints.
Maintaining historical layouts without manual schema migrations was solved by serializing complete canvas graphs to JSON in a versions table and calculating active schema differences using deep-diff logic.
Built an end-to-end machine learning-powered intrusion detection system (IDS) that trains, benchmarks, and ranks 7 classification algorithms (Logistic Regression, Decision Trees, Random Forests, XGBoost, SVM, KNN, and Gradient Boosting) to identify and block security anomalies in network session traffic.
7 Ensemble Models Benchmarked · Stratified Preprocessing · Precision & Recall Optimization
Designed a multi-stage data science pipeline: dynamic log ingestion via pandas, exploratory analysis (heatmaps, box plots), features preprocessing (StandardScaler and encoding), and stratified 80/20 train/test execution of 7 competitive models.
Prioritized Recall scoring over basic accuracy as false negatives represent undetected security breaches. Also implemented stratified splitting to ensure rare intrusion patterns are represented in both validation sets.
Handling highly imbalanced session repositories where 99% of logs are benign. Mitigated bias by optimizing class weights inside classification configurations and adjusting model decision threshold barriers.
Competitive results and milestones, most recent first.
Advising on strategic initiatives, tech community events, and collaborative engineering assemblies.
Coordinating between corporate recruiters and student candidates, driving campus placement logistics.
Managed operational frameworks, workshop schedules, and cross-university technology symposiums.
Fostering collaborative technical learning and representing external tech initiatives on campus.
Collaborated on system architecture, design specifications, and successful filings for a hardware-related utility patent.
Led planning and implementation of workshops, expert speaker panels, and diversity in engineering hackathons.
Recognized at the national level for designing high-performance cloud application architectures.
Awarded First Place for developing an innovative interactive simulation prototype within 48 hours.
A bit of background, plus where I've studied and worked.
I am a Computer Science and Engineering student at UPES Dehradun, specializing in CCVT (Cloud Computing & Virtualization Technology). My engineering approach emphasizes robust systems architecture, automation, and algorithmic problem-solving. I design clean pipelines, model real-world networks mathematically, and verify code quality with structured methodologies.
B.Tech – Computer Science and Engineering (CCVT)
Science & Mathematics
General Academics
Working on Agentic AI frameworks and integrating autonomous model systems for enterprise software modules. Developing intelligent automation agents and orchestrating LLM prompt routing APIs.
Successfully engineered and deployed the dynamic CRM form builder module for the company's client intake and workflow automation systems. Building robust data engineering pipelines, optimizing ETL operations, and database configurations.
Led software testing, UI/UX usability analysis, and cross-functional debugging. Managed client coordination for patent filings, supporting smooth onboarding.
Managed end-to-end client interactions and project lifecycle communication. Ensured prompt, structured engagement, streamlining multi-stakeholder workflows.
Assisted in requirements gathering, business process modeling, and functional specification drafting. Analyzed system usage data and user workflows to identify process bottleneck barriers.
Researched memristors for brain-inspired neuromorphic computing hardware. Contributed to ML-based modeling, circuit simulations, and hardware designs.
Tools and technologies I reach for daily, organized by where they sit in a system.
Credentials earned alongside coursework and project work.
AWS Training & Certification
Jun 2026
AWS Training & Certification
Jun 2026
AWS Training & Certification
May 2026
AWS Academy
Jan 2026
AWS Academy
Dec 2025
AWS
Sep 2025
Open to internships, engineering discussions, collaborations, or simply networking. Reach out directly or send a message.