Building scalable Generative AI systems on AWS.

Experience

A career built on shipping AI to production.

1. May 2026 — Present Vancouver, Canada

   Machine Learning Architect · Caylent

   Promoted to ML Architect; lead end-to-end architecture and delivery of scalable GenAI and ML solutions on AWS — from discovery through production rollout — with focus on cost-performance trade-offs, stakeholder alignment, and maintainable systems.

2. Dec 2023 — May 2026 Vancouver, Canada

   Senior Machine Learning Engineer · Caylent

   • Led architecture, delivery, and client engagement across 6 enterprise engagements (Fintech, Automotive, MarTech, MedTech, EdTech).
   • Designed GenAI systems on Amazon Bedrock: RAG pipelines, vector retrieval, agent-based architectures (ReAct, planner-executor, multi-agent).
   • Established LLM evaluation frameworks (text-to-SQL, RAG quality, agent reasoning) and LLMOps practices: prompt versioning, model evaluation pipelines, observability.
   • Co-authored Caylent's MLOps Solution Offering — adopted company-wide as a reference for client engagements.
   • Mentored 2 engineers on GenAI architecture and MLOps to independent ownership of client deliverables.

   Selected client outcomes

   Fintech

   97% faster financial statement processing

   GenAI document pipeline: 30–60 min → ~90 s; >95% extraction accuracy; 10× throughput (50→500 docs); $0.36–0.47 per document, enabling same-day reviews.

   MarTech / CRM

   Fine-tuned LLMs in production for 25,000+ orgs

   DistilBERT email classification (99.3% F1, 0.064s inference, $0.23/10k records) and Mistral-7B name extraction (97%+ accuracy) across beta, prod-us, prod-eu — among the first production deployments of fine-tuned Mistral-7B for enterprise NLP.

   Automotive

   Sole technical lead on multimodal AI avatar evaluation

   Benchmarked 5 models (OmniAvatar, HunyuanVideo, Wan2.2, Wan2.2-S2V-14B, Amazon Nova Reel) across quality, generation time (2–35 min/18s video), and cost ($1.38–$2.98/video) on AWS p5.4xlarge (H100 80GB). Customer verdict: "A great success. Work of the highest quality."

3. Sep 2022 — Jun 2023 Vancouver, Canada

   Senior AI Engineer · Alida Inc

   • Led development and productization of AI services, managing a team of 4 engineers.
   • Designed AI pipelines processing 15M+ records, securing a multi-year enterprise contract.
   • Built reusable text analytics frameworks (PII masking, sentiment, taxonomy) on Node.js + Python; scaled on Terraform + AWS EKS.
   • Managed MLOps for 13+ microservices — standardized CI/CD, testing, deployment.
   • Helped grow the AI engineering team from 2 → 6 via interviews and recruiting; mentored on NLP, MLOps, cloud-native AI.
4. Apr 2020 — Sep 2022 Vancouver, Canada

   AI/ML Engineer · Alida Inc

   • Designed, deployed, and maintained 5 core NLP microservices using spaCy, NLTK, IBM NLU, AWS Translate.
   • Built CI/CD with Terraform + GitHub Actions; deployed XGBoost ML pipelines to production via AWS EKS.
5. Jan 2014 — Aug 2019 Bangalore, India

   Software Engineer · Nagra Kudelski

   Grew from individual contributor to UI lead across 6+ projects serving 2M+ users (Chivas TV, Euskaltel TV, TBC Taiwan, FOXTEL Australia).

   • Coordinated 4 engineers across 3 concurrent projects; led requirements, sprints, release sign-off.
   • Key contributor at FOXTEL Australia: Cloud DVR, content protection, parental lock, GA-based recommendations.
   • Supported live 4K content during the 2016 Rio Olympics — resolved production issues on-ground in Brazil, securing a 2-year contract extension.
   • Mentored 6 engineers on UI architecture, code quality, and engineering practices.

Projects

Selected open-source & learning projects.

Live data from github.com/manju-malateshappa — refreshed each load.

See all repositories on GitHub →

Education

Academic foundation.