Experience | Lokesh Reddy V

EXPERIENCE / EDUCATION

Sep 2026 — Present

Tata Consultancy Services

Generative AI Engineer

Building an enterprise GenAI application platform for complex workflow automation, combining a modern web interface, FastAPI services, application data, retrieval pipelines, and multi-agent orchestration.

PROGRESSIONApplying earlier NLP and full-stack AI experience to controlled enterprise GenAI delivery.

Design LangGraph workflows with deterministic states, structured contracts, controlled tool execution, validation loops, and mandatory human-review gates for high-impact actions.
Build access-aware retrieval and evaluation capabilities with source attribution, output validation, audit logging, monitoring, versioning, and rollback controls.

Jan 2024 — Aug 2026

KPMG

AI Engineer — AI/ML & Agentic Systems

Led production AI and agentic systems for audit and advisory workflows with bounded orchestration, retrieval, evaluation gates, human review, confidentiality controls, and operational observability.

PROGRESSIONExpanded from full-stack AI delivery into regulated document intelligence, agent state, model adaptation, evaluation, and recovery design.

Reduced manual document-review effort by 34% and reduced evidence-search time from about 40 minutes to under 4 minutes through grounded audit intelligence.
Built 200+ domain evaluation cases, expanded coverage to 87%, blocked 11 regressions, improved task F1 from 0.74 to 0.91, and validated 10× traffic with Kubernetes autoscaling.

Related case study

Aug 2021 — Dec 2022

BT Group

Full Stack AI Engineer

Built retrieval-backed infrastructure guidance and natural-language automation capabilities supporting more than 100 engineers, with policy-aware templates and approval controls.

PROGRESSIONConnected model capability to usable services, retrieval, policy-aware templates, and developer workflows.

Improved infrastructure provisioning speed by 60% while reducing template errors by 74% through validated natural-language workflows.
Combined curated engineering documentation, source-aware responses, policy checks, and human approval in a developer-facing assistant.

Related case study

Nov 2018 — Dec 2019

Tech Mahindra

Application Development Engineer — AI/ML Practice

Developed BERT and RoBERTa-based document classification, sentiment analysis, and request-routing capabilities for enterprise applications.

PROGRESSIONStarted with measurable transformer-based NLP features and the application contracts around them.

Built a BERT document classifier reaching 93.4% macro F1 and reproducible training, error-analysis, threshold-tuning, and regression workflows.
Delivered model predictions through stable REST interfaces, reducing misrouted tickets by 31% with versioned MLflow patterns.