Solutions Architect at MiPhi Semiconductors (Micromax × Phison JV) · Bengaluru
Architecting on-prem and cloud AI platforms
AWS Certified

I turn AI and cloud ambitions into architecture that actually runs.

Aditya Jha · Cloud and AI Solutions Architect

Started as a platform engineer, now designing AI solutions end to end: requirements, architecture, sizing and delivery, with pre-sales and program management along the way.

View my experience Download resume

LinkedIn ↗GitHub ↗Medium ↗

$ whoamiAditya Jhasolutions-architect --ai --cloud$ kubectl get experienceMiPhi Semiconductors2026 → presentOsfin.ai2025 → 2026AMD India2024 → 2025$ cat stack.txtaws · k8s · terraform · ci/cdvllm · gpus · right-sizingai-infrastructure · pre-sales

About

I sit where engineering meets the customer. I started with LLM fine-tuning at AMD, then ran production EKS clusters on AWS with Terraform, Helm and ArgoCD. Today I architect AI solutions end to end: I turn a client's requirements into a sized, costed design, write the proposal, and see it through delivery.

I've worked with universities and enterprises and supported defence and government tenders, on everything from coding-agent platforms to multi-node GPU clusters. I care about designs that are right-sized, explainable and ready to build.

Experience

Solutions Architect (Pre-Sales) & Program Manager
MiPhi Semiconductors · Phison × Micromax JV
Feb 2026 → Present
- Closed a ₹2 Cr+ Kubernetes GPU cluster order with an education customer, via a channel partner, within 7 months of joining. Delivered the BoQ, power and cooling sizing, and licensing, SLA and security responses.
- Lead solutions architect for the team's AI deals across 10+ enterprise, education and public-sector engagements, including an 18,000-user on-prem coding-agent proposal to replace a cloud LLM API bill.
- Appointed project manager of a 5-person agentic AI product team, running scrums, Jira tracking and weekly management updates.
- Program manager for a multi-party server program with a government R&D body and two industry partners.
- Audited a GPU fleet capacity model, fixed 4 modelling errors and rebuilt it across 40 model and precision combinations.
- Led a 12-deliverable go-to-market package for an on-prem agentic AI appliance, from sales deck to TCO model and user manual.
AI solution architectureLLM inference sizingGPU clustersTCO modellingProgram management
DevOps & Platform Engineer
Osfin.ai
Jun 2025 → Jan 2026
- Built and secured production Amazon EKS clusters with modular Terraform (IRSA, managed node groups, workload isolation), cutting cluster provisioning time by 40%.
- Fixed CI/CD and Helm pipeline failures across environments, cutting build failures by 30%.
- Automated multi-cloud operations on AWS, OCI and Azure with Jenkins, n8n and Python, cutting compute cost by 60% and saving 6+ engineer-hours a week.
Amazon EKSTerraformHelmArgoCDJenkinsFinOps
Co-op Intern, GenAI
AMD India
Jul 2024 → Jun 2025
- Fine-tuned Indic-language LLMs with INT8, INT4 and GPTQ quantisation for about 20% better inference efficiency on low-resource GPUs.
- Configured and deployed an end-to-end fine-tuning suite with automated inference, evaluation and benchmarking, orchestrated on Kubernetes.
LLM fine-tuningQuantisationKubernetesMLOps
Trainee
Corizo
May 2023 → Jul 2023
- Designed a highly available multi-tier AWS architecture (EC2, RDS, S3, auto scaling, load balancing) with Lambda automation and CloudWatch alerting.
AWSServerlessMonitoring