Waseem Akram
Staff Software Engineer — Systems & Applied AI
I build backend infrastructure, developer tooling, and applied AI systems.
CTO at Charing Cross Capital. Creator of HyperDjango. Based in Chennai, India.
02 / Selected Work
Server-driven hypermedia framework for Django featuring SvelteKit-style file-based routing, native SSE streaming, and automated Vite asset bundling.
Rezolution Engine
Automated legal document ingestion, vision extraction, and litigation monitoring platform for insolvency proceedings.
Optik & ProcureMesh
Touchless accounts payable automation and universal cXML PunchOut gateway for the SAP Ariba network.
SmoothPay
Enterprise invoice reconciliation engine with multi-tier approval workflows and cryptographic PDF signing.
03 / Systems & Craft
Principles
Server-driven over client state bloat
Prefer hypermedia, server-rendered components, and lightweight streaming over complex client-side state machines whenever possible.
Filter early, infer late
Optimize expensive model inference by filtering, pruning, and structuring data with deterministic and on-device tools before hitting large cloud models.
Declarative and reproducible
Treat developer environments, system packages, and deployments as code through Nix and declarative configs—eliminating "works on my machine" drift.
Boring foundations, sharp tools
Build on proven, rock-solid backends (PostgreSQL, Django, Linux) while staying aggressive and curious with modern developer tooling and applied AI.
Production Stack
Languages & Frameworks
Python, Django, Celery, Redis, PostgreSQL, pgvector, REST APIs, Server-Sent Events (SSE).
Developer Environment
Terminal-native workflow in Neovim (custom Lua configuration), tmux session multiplexing, Linux internals, and declarative system management via Nix and nix-darwin.
Infrastructure & Cloud
Docker Compose, Caddy (automated TLS/SSL reverse proxy), SAP BTP Kyma (Kubernetes enterprise tenant isolation), AWS, Trivy container security scanning.
Foundations & Learning
First-Principles Deep Learning
Mechanistic study of neural network gradient flows, backpropagation, and transformer architectures built from scratch with multi-head self-attention (following Andrej Karpathy's foundational series).
Fine-Tuning & Parameter Efficiency
Theoretical and practical mechanics of PEFT—low-rank adaptation (LoRA, QLoRA rank decomposition), prefix tuning, and quantization techniques.
Alignment & Inference Engineering
Preference optimization (DPO, RLHF), KV-cache optimization, speculative decoding, and local on-device model execution.
Document & Entity Intelligence
Production Vision LLM prompting, layout evaluation with Tesseract OCR, and zero-shot Named Entity Recognition with GLiNER.
04 / Experience & Resume
Detailed metrics, project histories, and architectural breakdowns are documented in my resume.
Charing Cross Capital
Chief Technology Officer (CTO)
Waak Labs
Founder & Lead Developer
Laagom
Lead Software Engineer
Early Native Mobile
iOS App Developer (Swift, Objective-C)
Education
B.S.A Crescent Institute — B.Tech in Computer Science & Engineering · GPA: 8.7 / 10.0