Home/About & Leadership

Engineering Led by Principals, Zero Outsourcing Bloat

MacroData Systems is built on an uncompromising principle: enterprise-grade software should be architected and written by seasoned engineers, not brokered through account managers or hidden subcontractors.

ORGANIZATION & LEADERSHIP

Founder-Led Engineering, Zero Agency Bloat

MacroData Systems is a registered micro-enterprise operated by Harsh Raj Singh out of India and London, UK. We eliminate layers of account managers, sales retainers, and junior outsourcing pools so you work directly with the engineer architecting and shipping your product.

Harsh Raj Singh

Founder & Principal Systems Engineer

London, UK

Senior full-stack & distributed systems engineer with 7+ years shipping high-throughput production backends across Tata Consultancy Services (TCS Digital Cadre) and Capgemini Engineering. MSc in Advanced Computer Science from Queen Mary University of London (specializing in Neural Networks, Deep Learning, Quantitative Risk & Systems Architecture).

Verified Career Track Record (LinkedIn):

Academic Demonstrator (QMUL): Instructed cohorts in Neural Networks & Deep Learning, hands-on model training, loss optimization, and neural architectures.
Engineering Squad Lead (Capgemini): Directed full-stack squads architecting the Number Reputation Management platform (high-concurrency Java / Spring Boot microservices).
Top 1% TCS Digital Cadre: Competitive selection into TCS Digital engineering group; designed enterprise recommendation engines and low-latency stream processors.
Published Systems Creator: Engineered Lumina (multi-LLM client-side RAG copilot on Chrome Web Store) and Auto-Echo CPU cache topology latency research.

Education

MSc Adv CS (QMUL, London)

Background

Ex-Capgemini & TCS Digital

Enterprise Type

Micro (Udyam Registered)

Udyam Certificate

UDYAM-UP-01-0216475

LinkedIn Profile (singhharsh199) ↗

Commercial Transparency

How We Partner With You

No Fabricated Track Records or Fake Logos

We do not invent client rosters or claim held corporate security certifications. We describe exactly what architectures we can build and provide verifiable reference blueprints.

Milestone-Based Delivery

Projects are divided into transparent technical milestones. You inspect working code and functional previews before releasing subsequent phase payments.

100% Client IP Ownership

All custom repository code, infrastructure configurations, and database schemas are handed over to your organization upon payment. Zero proprietary lock-in.

Specialists Scoped Per Project

Harsh leads engineering and architecture directly. For specialized domains (such as penetration testing or native mobile driver modules), vetted specialists are contracted per project.

Verified Career Track Record

7+ Years in High-Concurrency Systems

2025 – 2026 • London, United Kingdom

Queen Mary University of London (QMUL)

MSc in Advanced Computer Science. Led cohorts in neural network architectures, model training, loss optimization, backpropagation, and hyperparameter tuning. Engineered and benchmarked machine learning and deep learning projects including diagnostic predictive classifiers (Logistic Regression, KNN), client-side RAG vector embeddings, and unsupervised timing analysis models.

2023 – 2025 • Gurugram, India

Software Engineer Lead • Capgemini Engineering

Led engineering teams architecting the Number Reputation Management platform. Shipped high-throughput Java/Spring Boot microservices detecting spam numbers and automating carrier-grade remediation.

2018 – 2023 • New Delhi, India

Systems Engineer • Tata Consultancy Services (TCS)

Selected into the competitive TCS Digital Cadre (Top 1% engineers in AI/ML). Directed squads of multiple developers building enterprise recommender systems, stream processors, and internal analytics engines.

Innovations & Open Engineering

Featured AI & Systems Research

Auto-Echo • Side-Channel Latency AnalysisSystems Research

Unsupervised CPU cache topology mapping and side-channel detection on modern Apple M1 and Intel multi-core platforms.

Lumina • Multi-LLM Context AI CopilotChrome Web Store

Multi-model AI Copilot unifying GPT-4, Claude 3.5, and Gemini with fully client-side RAG and zero-leakage vector matching.