🚀 Senior Software Engineer @ Deliveroo (DoorDash)
💻 Distributed Systems · ML Systems · AI Agents
📍 London, UK

🔗 LinkedIn | Personal GitHub | Work GitHub

• 🛒 Deliveroo Express — Recently transitioned to a new vertical domain focusing on rapid delivery systems and scaling new product lines
• 🧠 Atlas — Building an AI-powered system (agents + knowledge + automation)
  👉 https://github.com/gagansingh894/atlas

I build high-scale backend systems and ML platforms that operate in production with real-world impact.

• ⚡ Systems processing 30K+ events/sec
• 🤖 ML serving at 1000 RPS, sub-20ms p99 latency
• 💰 Delivered systems driving £100M+ annual revenue
• 🧩 Focus on scalability, performance, and clean architecture

Ad Platform · Financial Systems · MLOps · Deliveroo Express

• 📈 Scaled Ad Platform managing 35,000+ campaigns → £100M+ revenue
• 🎯 Built targeting & scheduling systems → +10% adoption, improved ROI
• 💸 Delivered Ad Credits & Co-funded Campaigns → +£3.5M revenue
• 🧾 Built centralized billing system → £1.92M operational efficiency
• 📊 Enabled ads in filtered lists & offers → +£2.5M uplift
• 🔄 Led decoupling of financial systems via async/event-driven architecture
• 🤖 Built low-latency ML serving platform
  • 1000 RPS, sub-20ms p99
  • A/B experimentation + full MLOps lifecycle
• 🧠 Built LLM-powered campaign management agent (PoC)
• 🛒 Recently transitioned to Deliveroo Express vertical

Thousense: Cloud-Agnostic No-Code Data Science Platform

• 🏗️ Designed core architecture for data & ML pipelines (Python OOP + functional design)
• ⚙️ Built distributed execution engine using:
  • Celery · RabbitMQ · Dask · Kubernetes
  • Enabled concurrent scheduling & on-demand execution at scale
• 🔁 Developed hyperparameter tuning framework:
  • Optuna + Dask + Joblib → scalable parallel optimization
• 🌐 Built FastAPI-based backend APIs powering multiple microservices

• 📈 Built time-series forecasting models (Prophet, Statsmodels) → ~10% MAPE
• ⚡ Reduced 30% development time via reusable ML lifecycle framework (MLflow)
• 📉 Improved prediction accuracy by 5% via forward feature selection
• 🔄 Processed large datasets via PySpark (Hive → RDS pipelines)
• 🚀 Reduced 50% training time by orchestrating 5000+ pipelines using Dagster

• 🔍 Built framework to identify key market drivers
  • Combined PCMCI causality + SHAP explainability
• 🧪 Developed What-If simulation tool
  • FastAPI + React + RDS + S3
  • Enabled business users to explore scenario outcomes
• 🤝 Worked cross-functionally to deliver data-driven insights

• 🛠️ Built end-to-end ML pipeline (Scikit-learn, Optuna)
• ☁️ Deployed via Argo on Kubernetes (AKS)
• ⚡ Reduced 40% training time using:
  • Containerized workloads (Go + MongoDB)
  • Parallel execution of 1000+ training jobs

Languages
Go Rust Python

Backend & Systems
Distributed Systems · Event-Driven Architecture · DDD · Microservices · gRPC · REST

AI / MLOps
MLflow · Airflow · Argo · LangChain · RAG · AI Agents

Data & Storage
Kafka · Postgres · DynamoDB · Redis · Cassandra · Qdrant

Cloud & Infra
AWS · Azure · Kubernetes · Docker · Terraform · Datadog · CircleCI

• High-performance, cloud-agnostic ML serving system
• Rust-based → low latency, high throughput
• Multi-framework + OpenTelemetry observability

• Hybrid recommender + conversational AI
• Full-stack system (Python + Rust + React)
• End-to-end MLOps lifecycle

• Streaming ML system using Spark
• Real-time anomaly detection + ETL pipelines

• Event-driven architecture
• ACID-compliant ledger + reconciliation
• Designed for fault tolerance

• Distributed systems & system design
• ML infrastructure & production AI
• LLMs, agents, and RAG systems
• High-performance backend systems (Rust/Go)

• Most production systems are private (Deliveroo & Thoucentric)
• Happy to discuss architecture, trade-offs, and scaling decisions

If you're working on:

• Distributed systems at scale
• ML platforms / AI agents
• AdTech / FinTech / Data platforms

👉 Let’s connect