8 min read2024
Building a Real-Time F1 Strategy Engine
How I architected a low-latency prediction system for Formula 1 race strategy using streaming telemetry data and ensemble ML models.
ML SystemsReal-TimeF1
Available for opportunities
Alan
Subhash
AI EngineerMachine LearningData Systems
Building intelligent systems that transform real-world data into predictive insights.
8+
AI Projects
5+
ML Models
15+
Technologies
Scroll to explore
Portfolio
Intelligent Systems I've Built
Each project represents a complete AI engineering solution — from data ingestion to production deployment.
Featured
Watch Demo
MindAI – Mental Health Support
AI-powered mental health support application combining NLP and reinforcement learning. Uses DistilRoBERTa for real-time emotion detection across 7 emotions, and a Dueling Double DQN agent to learn user preferences and deliver personalized content recommendations including breathing exercises and motivational content.
FastAPIReactPyTorchDistilRoBERTaDQN
View on GitHub
SmartSense Hood (V-Guard)
Engineered a ductless exhaust hood to improve air quality and energy efficiency. Integrated a whistle counter for monitoring pressure cookers using sound and temperature sensors. Features included LPG leak detection, battery backup, and Wi-Fi alerts to smartphones for real-time safety monitoring.
ArduinoIoTFirebaseMLSensors
Drought Prediction Model
Expanded a machine learning model for drought prediction using ETL processes in Python. Performed preprocessing, feature engineering, and data visualization using Matplotlib to improve model accuracy to 85% and uncover trends relevant to climate forecasting.
PythonScikit-learnPandasETLMatplotlib
Fire Fighting Robot
Developed a Python-based autonomous robot capable of detecting fire zones and deploying CO₂ balls. Incorporated real-time sensor data validation and path optimization using regression models to enhance hazard detection and navigation.
PythonSensorsRegressionRobotics
AeroBot X
Enhanced an AI-powered air-purifying robot with HEPA filters and dynamic air quality monitoring. Focused on energy efficiency, achieving 20% battery optimization using path planning algorithms and onboard sensor integration for real-time decision making.
PythonROSOpenCVRaspberry PiSLAM
F1 Tyre Degradation Prediction
Developed a regression model to forecast Formula 1 tire degradation and optimize pit stop strategies. Analyzed race data and visualized patterns with Python and Matplotlib to support competitive decision-making during races.
PythonTensorFlowFastAPID3.jsMatplotlib
Self-Driving AI Application (IHRD)
Created a prototype of a self-driving vehicle using Python and sensor data. Integrated real-time object detection and feature extraction pipelines. Used visualizations to evaluate efficiency and connect results to AI-based navigation strategies.
PythonOpenCVDeep LearningSensors
Tinku – Mood-Based Media App
Developed a sentiment-aware web application that recommends YouTube and Spotify content based on user emotion. Built RESTful APIs, used SQLite for data management, and designed the interface using HTML, CSS, and JavaScript. Deployed to the cloud via Render.
PythonFlaskSQLiteREST APINLP
Deep Dives
Engineering Case Studies
Detailed breakdowns of end-to-end AI system architectures, from data ingestion to production deployment.
Case Study
F1Track.AI
Real-Time Race Analytics Platform
Problem
Formula 1 teams need rapid strategy decisions based on live telemetry. Existing tools lack real-time predictive capability for mid-race adjustments.
Data Sources
- Live telemetry feeds
- Historical race data
- Weather APIs
- Tire compound datasets
Architecture Pipeline
Data Sources
Telemetry, weather, tire data
ETL Pipeline
Real-time stream processing
Feature Engineering
Lap time deltas, tire wear curves
ML Models
XGBoost + LSTM ensemble
Prediction Engine
Strategy simulation engine
Dashboard
React + D3.js visualization
Results
92%
strategy prediction accuracy
Sub-second
latency on predictions
Processed
50K+ telemetry events per race
Case Study
Drought Prediction AI
Environmental Risk Forecasting System
Problem
Agricultural regions lack early warning systems for drought events. Traditional forecasting relies on manual analysis of limited weather data.
Data Sources
- Satellite imagery
- Weather station data
- Soil moisture sensors
- Historical drought indices
Architecture Pipeline
Data Sources
Satellite, weather, soil data
ETL Pipeline
Batch + streaming processing
Feature Engineering
SPI, NDVI, temporal features
ML Models
Random Forest + Gradient Boosting
Prediction Engine
Risk probability scoring
Dashboard
Geospatial risk heatmap
Results
87%
prediction accuracy 30 days out
Coverage
across 5 agricultural zones
Reduced
false alarm rate by 40%
Interactive
Try My AI Models
Experience live demonstrations of machine learning models — interact with real prediction engines.
Emotion Detection AI
NLP-powered sentiment analysis
Drought Risk Predictor
Environmental ML model
50
30
40
System Design
AI System Architecture
Explore the end-to-end pipeline that powers intelligent systems — hover over each component to learn more.
Data Ingestion
Kafka / REST APIs / WebSockets / S3
Data Processing
Apache Spark / Pandas / dbt / Airflow
Feature Engineering
Feature Store / Scikit-learn / NumPy
ML Models
TensorFlow / PyTorch / XGBoost / MLflow
Prediction Engine
FastAPI / TensorFlow Serving / Redis
Visualization
React / D3.js / Grafana / Plotly
Hover over a pipeline component to explore its details
Capabilities
Technology Ecosystem
A comprehensive map of the tools, frameworks, and platforms I use to build intelligent systems.
Machine Learning
Python95%
TensorFlow85%
PyTorch80%
Scikit-learn90%
Keras85%
Feature Engineering85%
Data Engineering
SQL90%
ETL Pipelines85%
Apache Hadoop70%
Distributed Computing70%
Pandas95%
Data Cleaning90%
Cloud & DevOps
GCP80%
AWS75%
Azure70%
Docker80%
REST APIs90%
Firebase85%
Development
React85%
FastAPI85%
Flask80%
Java70%
Git / GitHub90%
R65%
Credentials
Certifications
Professional certifications and credentials in AI, cybersecurity, and technology.
Recognition
Hackathons & Achievements
Competitive milestones and recognition in AI engineering and innovation challenges.
2024Top 5
V-Guard Big Idea 2024
Top 5 Finalist among 300+ teams for AI-based IoT safety innovation with the SmartSense Kitchen Hood.
2024Winner
Hack for Humanity
Winner for building a data-driven drought prediction system achieving 85% accuracy.
2024Certified
Google Cybersecurity Foundations
Completed Google's professional cybersecurity foundations certification program.
2024Certified
Google AI Essentials & IBM SkillsBuild
Earned certifications in AI Essentials from Google and completed IBM SkillsBuild program.
Visual Archive
Project Gallery
Behind-the-scenes photos and demo videos from building AI systems and competing at hackathons.
Demo Videos
Project Demo
Project Demo
Project Demo
Project Demo
Fire Fighting Robot
Tinku App Demo
Insights
AI Engineering Insights
Technical writings on machine learning systems, data engineering, and building AI at scale.
12 min read2024
Machine Learning for Environmental Forecasting
Deep dive into building drought prediction models with satellite imagery, weather data, and feature engineering techniques.
Environmental AIFeature Engineering
10 min read2024
Designing Scalable AI Systems
Principles and patterns for building production-grade ML pipelines that scale from prototype to deployment.
ArchitectureMLOpsScale
Background
About Me
Artificial Intelligence and Data Science undergraduate specialising in machine learning, predictive analytics, data engineering, and AI-driven system design. Experienced in building scalable ML pipelines, optimising real-time intelligent systems, and deploying cloud-based analytics solutions.
Strong expertise in Python, SQL, ETL workflows, and model evaluation techniques. From robotics to NLP-powered applications, I build production-grade AI systems that solve practical problems.
6+
Languages
8+
Projects
AI/ML
Focus
Experience & Education
2022 - 2026 (Expected)MITS, Kochi
B.Tech in AI & Data Science
Specializing in machine learning, deep learning, and data engineering with focus on practical AI applications at Muthoot Institute of Technology and Science.
Jan 2025 - Mar 2025Robotics & AI
AeroBot X – Autonomous Air Quality System
Engineered an AI-powered autonomous air quality monitoring robot. Improved navigation efficiency by 20% and reduced pollutant detection latency by 30%.
2023 - PresentIndependent Projects
AI Research & Development
Building production-grade AI systems including race analytics, environmental forecasting, mental health support, and IoT safety platforms.