Neural networks
CNNs, transformers, and custom heads — trained, fine-tuned, and shipped with eval suites.
Shankar Teja Lingala · AI Systems Engineer
Production AI built on Claude, Gemini, and ChatGPT. Deep learning, CNNs, RAG, multi-agent systems — and quiet curiosity for quantum and neural-tech frontiers. Remote from London, working anywhere.
scroll
01About
a life in three chapters.
Born in Eluru, trained in Vadodara, building from London. AI engineer fluent in Claude, Gemini, and ChatGPT — and the careful half of shipping software.
Eluru, Andhra Pradesh — a small town that taught me to be patient with hard problems. I picked up programming on a borrowed laptop and a slow connection. The constraint made me careful, and careful made me curious.
Six years across hospitality, retail, kitchens, and customer-service desks paid the bills between three degrees in three cities. The non-AI work taught me the soft half of building software — listening, reading rooms, and finishing under pressure.
0+
Years coding
0
Degrees · 3 cities
0
Languages spoken
0h
Reply window
02Capabilities
six things I actually do well.
Transformer attention
Multi-head attention, KV caching, and prompt design that respects the actual mechanism.
Embedding spaces
Vector search, semantic clustering, and grounded retrieval that doesn't hallucinate.
Training loops
Optimizers, schedulers, mixed precision — the careful side of getting models to converge.
Tool-using agents
Function calling, structured output, and clean failure modes — agents that can be debugged.
Biosignals & vision
EEG decoding and computer vision — CNNs and transformers working on the messy real world.
Drag right to explore
03Services
hand-built craft at production scale.
From a single CNN classifier to a full LLM-powered platform — a complete range of AI work for individuals, startups, and research teams.
/ 01
Deep learning models
CNN architectures and supervised learning pipelines for vision, classification, and structured data — from dataset curation to evaluation harness.
/ 02
Full-stack AI products
Production-grade applications powered by Claude, Gemini, or ChatGPT. Chat interfaces, internal tools, and dashboards that ship — and stay shipped.
/ 03
RAG & knowledge systems
Embeddings, vector search, and grounded generation — turning private documents and structured data into useful, accurate answers.
/ 04
Multi-agent workflows
Tool-using agents, planners, and coordinated pipelines. Built with function calling, structured output, and clean failure modes.
/ 05
Computer vision
Object detection, classification, and image-based ML — including medical and domain-specific imaging trained with PyTorch and OpenCV.
/ 06
Prompt engineering
Carefully designed prompts, evaluation harnesses, and Custom GPTs that behave consistently in production — not just in demos.
·Live
how it thinks.
A small live demo. Scroll past this section and watch a real prompt run through a Claude-style reasoning loop, character by character.
04Research frontiers
where AI meets the frontier.
Beyond the day-to-day LLM and CNN work, two domains pull my curiosity hardest. They sit at the edge of what's currently possible — which is precisely why they're worth learning now.
Quantum Computing
Variational quantum circuits, quantum kernels, and hybrid classical–quantum models. Learning Qiskit and PennyLane to bring quantum primitives into AI pipelines.
- /01Variational Quantum Circuits (VQC)
- /02Quantum kernels for classification
- /03Hybrid classical–quantum networks
- /04Qiskit · PennyLane · Cirq
Neural Technology
Where AI meets the nervous system — neural-signal decoding, brain-computer interfaces, EEG classification, and biosignal denoising with CNNs and transformer models. The frontier I'm most curious about.
- /01EEG signal classification
- /02Brain-computer interface research
- /03Biosignal denoising & features
- /04MNE-Python · OpenBCI · PyTorch
"I'd rather ship one careful thing than seven loud demos that break by Tuesday. The frontier rewards patience — and the engineers who can be honest about uncertainty."
— Shankar Teja Lingala · Working manifesto
05Selected work
the work speaks for itself.
A growing catalogue of AI projects — full-stack apps, deep-learning research, and quietly useful tooling. Each one shipped end-to-end.
Trident AI
A personal AI chat product I built end-to-end — animated branding, in-browser code execution, persistent conversations.
Case study →Gastro Vision
A multi-class CNN classifier for gastrointestinal disease detection, built for the CN7023 AI & Machine Vision module.
Case study →Portfolio Engine
A Python toolkit that produces tailored, role-specific CVs and cover letters from structured config. Same source, dozens of tuned outputs.
Case study →Studio ✦
An ongoing prototype that blends applied ML with a domain I'm self-teaching. I'm keeping the details quiet until it ships.
Coming soon →Eval & Data
Reusable evaluation harness and dataset-preparation pipeline I rely on across CNN experiments. One config, full eval.
Case study →Custom GPTs
Domain-specific Custom GPTs with function calling and structured output, built for repetitive workflows that punish brittle prompts.
Case study →·Process
a path without friction.
From first message to final hand-over, every step is small, clear, and on the table.
/01 — DISCOVERY
First contact
We start with a conversation — your problem, your data, your constraints. I respond within 24 hours with a clear scope.
→/02 — BRIEF
Quote & plan
Detailed evaluation of approach, model choice, and timeline. Transparent, fixed-scope quote within 48 hours.
→/03 — BUILD
Iterate fast
I prototype fast, iterate on real data, and keep you in the loop with weekly walkthroughs. No black boxes.
→/04 — SHIP
Deliver & support
Final review, documented handover, and ongoing support. The repository, the prompts, the eval suite — all yours.
·Education
from eluru to london.
Three programmes, three cities, one through-line — building toward production AI.
MSc Artificial Intelligence
University of East London — London, UK
Frontier AI · LLM systems · Applied ML research
B.Tech in Artificial Intelligence
Parul University — Vadodara, India
Machine Learning · Deep Learning · Computer Vision
Diploma in Computer Science
CRR Reddy College — Eluru, India
Foundations · Programming · Data Structures