dote
Vigneshwaran Siva Sankaran
// ML / AI Engineer Data Scientist Researcher Esports Player  — building things that matter._
📍 La Jolla, CA 🎓 UCSD MSBA · Dec 2026 🔗 itsdote.tech
About · hover & throw the letters
Overview
💼RSKD Labs · Founding Engineer
🎓UC San Diego · M.S. Business Analytics
📊IIT Madras · Data Science & Programming
🏛️SASTRA University · B.Eng. Biotechnology (B.Tech)
📍La Jolla, California
✉️hi@itsdote.tech
Technical Skills
Languages
PythonSQL JavaC/C++ JavaScriptBash
ML / AI
PyTorchTensorFlow scikit-learnRAG LSTMCNN LLM Agents
Data
PySparkPandas PolarsNumPy Hadoop
Infrastructure
AWSDocker KubernetesCI/CD ETL PipelinesLinux
Experience
Jul 2023 – Jun 2025 · Chennai, India
RSKD Labs
Founding Engineer · AI Systems & Infrastructure
−30%pipeline latency
+25%production value
30 FPSCV inference
600mse2e latency
10+components
800krecords/run
🛸 AI Drone Security System

Full ML stack for a security-critical autonomous drone platform — RAG, LLM reasoning, CV inference. Owned all MLOps on AWS with Dockerized CI/CD and automated retraining pipelines.

Jan – May 2023 · Kattankulathur, India
Zoho Corporation
Member Technical Staff Intern · Zoho Payroll
1M+users
−15%SQL exec time
5s→2sload latency

Production backend APIs for Zoho Payroll processing tens of millions of records in a high-availability financial system. Deep SQL optimization under transactional load.

Project · Feb 2026
01
SkillRank Lab — Skill Rating & Matchmaking
Python · Elo · Glicko-2 · CSE · DualMMR · scikit-learn
🎮 99.5% variance reduction · 80K+ matches

Competitive matchmaking research system comparing 4 rating algorithms across calibration & SBMM simulation. Built directly in response to Riot Games' Senior DS (Skill & Matchmaking) role.

Scale500 calibration · 5,000 SBMM · 80K+ matches · 60-day sim
CSE Brier0.2508 (near theoretical min 0.25) — beats Glicko-2 by 13%
Spearman0.28 → 0.57 rank correlation over 60-day window
Smurf detectF1=0.50 · 45% recall · 56% precision · 4 behavioral signals · day-7
Project · Aug 2025
02
VALORANT Player Intelligence
scikit-learn · Logistic Regression · Gradient Boosting · Streamlit · Henrik API
📊 ROC AUC 0.9949 pro · 0.9885 public

Behavioral segmentation + win probability modeling across pro VCT and public VALORANT cohorts. Real match data from Henrik API — not synthetic.

Data4,211 pro VCT + 1,347 public competitive matches — real data
Pro modelLogistic Regression · ROC AUC 0.9949
Public modelHist Gradient Boosting · ROC AUC 0.9885
AnalysisRole-conditioned clustering · agent-behavior drift · archetype ID
🏆 SDx Hackathon Winner · Jan 2026
03
AI-IaSS — Cloud Infra Cost & Security Optimizer
LLM Agents (Gemini) · Python · SQL · AWS · GCP · Azure
💰 20–30% cost reduction · 100+ configs evaluated

Agentic system translating natural language infra requirements into cloud architecture recommendations. Evaluates configs via integrated cost modeling + 50+ security rules across AWS, GCP, Azure.

Award🏆 SDx Hackathon — Best Technical Track, Jan 2026
Scale100+ candidate configs per request · 50+ integrated security rules
ImpactEstimated 20–30% cost reduction per recommendation
DeployedLive at masda.itsdote.tech
Project · Jul 2025 · APalaniaLab
04
RiboSense — Riboswitch Detection & Classification
TensorFlow · Google TPU (64-core) · CD-HIT · Mixed Precision
📈 0.98 AUC-ROC · 150% throughput · ~5× epoch speed

Deep learning pipeline for detection and classification of riboswitches in genomic sequences. Large-scale training optimization on 64-core TPU with rigorous genomic data deduplication.

Throughput+150% via TPU batch/input-pipeline tuning · epoch time ~5× faster
Preprocessing−40% time · CD-HIT dedup at 0.96 similarity · parallel preprocessing
Data hygieneSequence-level dedup before split — zero data leakage
Project · Feb 2026
05
Cache-Aware Routing for Distributed Inference
Python · Distributed Systems · Benchmarking
⚡ 12.6× avg latency · 100% cache hit efficiency

Cache-aware request routing system for multi-worker ML inference. Hash and heuristic policies benchmarked against random baseline under repeated-query workloads.

Latency12.6× avg improvement vs random routing
Cache100% hit efficiency under repeated-query workloads
AnalysisQuantified hit-rate vs latency trade-offs across routing policies
Project · Dec 2025 · UCSD MGTA-452
06
Aviation Analytics — Delay & Turbulence Intelligence
Python · ARIMA · scikit-learn · Plotly · Geospatial · Streamlit
✈️ 516K PIREP reports · 311 airports · BTS real data

Multi-module aviation dashboard: delay forecasting, turbulence prediction, airport efficiency, airline comparison using real BTS + PIREP data. Team: Vigneshwaran Siva Sankaran, Jing Luo, Vivian Yang.

DataBTS On-Time + T-100 Traffic · 516,665 PIREP reports · 311 airports
ModelsARIMA delay · airport efficiency regression · turbulence risk classifier
Viz3D geospatial heatmap · radar chart · KPI heatmap · seasonal trends
Education
2019 – 2023
SASTRA University
B.Eng. Biotechnology (B.Tech)
Minor in AI · Thanjavur, India
2020 – 2022
IIT Madras
Diploma, Data Science & Programming
Concurrent with SASTRA
2025 – 2026
UC San Diego
M.S. Business Analytics
ML · Statistical Modeling · La Jolla, CA
Get in touch
Let's build something fast, scalable, and real.
When I'm not coding
VALORANT — Diamond II · 5 stacks · entry/duelist
CS2 — FPS for 12+ years · still grinding ranked
Chess — currently losing to 1200 bots · Sicilian enjoyer