Senior Director, Operations & Independent Builder

Akhil Jaitly

I run AWS, network, datacenter, and systems teams at enterprise scale, and I build production software on the side: AI agent tooling, contract intelligence, and personal-data platforms. Source stays private. The work below is deployed and running.

Capacity planning · Incident response · Cost optimization akhiljaitly@gmail.com
6
Products shipped
4
Live in production
2
Patent-pending
IaC
Ansible + Terraform ops

Selected Work / 06

Verdict

Patent-pending

LLM agent observability platform. Captures agent trajectories, embeds and clusters them, and replays runs deterministically to catch regressions. Deployed to production.

FastAPINext.jsPostgresSDK (py/ts)
verdict.jaitly.dev · demo on request

ClauseWatch

In build

AI contract intelligence. Extracts license terms from PDF/DOCX with every value pinned to its source span and verbatim quote, confidence-gated to a human review queue, persisted as a constraint graph.

FastAPIClaudeSQLAlchemyNext.js
Private build

FamilyVault

Live

Family document vault: identity, medical, legal, and financial records with expiry alerts and member sharing. SSO across a shared identity tenant, encrypted blob storage.

FastAPIR2Logto SSOResend
family.jaitly.dev · demo on request

PropertyVault

Live

Property management: maintenance, expenses, income, documents, vendors, and reminders. Direct JWKS token validation, dashboard rollups across every property.

FastAPINext.jsPostgres
property.jaitly.dev · demo on request

FinanceHub

Live

Personal finance: accounts, transactions, budgets, bills, and savings goals with AI-generated insights. Imports transaction data and syncs with the property platform.

FastAPIClaudeNext.js
finance.jaitly.dev · demo on request

ARIA

Patent POC

Enterprise risk intelligence: unified engines for anomaly, capacity, and vendor-risk scoring under one platform. Proof of concept behind an active patent effort.

FastAPIPostgres 16Redis
Private POC

Infrastructure & Automation / ops

System Builds

Private

Ansible automation for standardized server build workflows: provisions VMs on VMware vSphere from template through OS configuration, repeatable and idempotent across tiers, driven by templated host and group variables. A new host comes up the same way every time, no manual steps.

AnsibleVMware vSphereVM provisioningJinja2Idempotent
Sanitized · patterns on request

Network Patching

Private

Ansible automation for network device patching: staged rollout, automatic config backup before change, and validation after, so a patch cycle that used to be manual and risky runs the same way every time and proves itself when it lands.

AnsibleNetwork devicesBackupValidation
Sanitized · patterns on request

Network Health Checks

Private

Ansible-driven health checks across network devices: templated probes gather state, flag drift and failures, and roll up into readable reports. Catch problems on a schedule instead of during an incident.

AnsibleJinja2Health checksReporting
Sanitized · patterns on request

Network Inventory

Private

Ansible automation that keeps a live, accurate inventory of network devices: collects facts from the fleet and renders them into a single source of truth, instead of a spreadsheet that goes stale the day it is written.

AnsibleInventoryFact-gatheringJinja2
Sanitized · patterns on request

AWS Terraform Labs

In build

A hands-on AWS lab built entirely in Terraform and documented as I go: reproducible VPC, compute, storage, and IAM scenarios mapped to the Solutions Architect Associate domains. Infrastructure as code, torn down and rebuilt on demand.

TerraformAWSHCLSA Associate
Docs & walkthrough on request

How I Build

Ship to production

Every project above runs on real infra with custom domains, SSO, and CI. Not demos.

Provenance first

Extracted or inferred data carries its source. Confidence gates route the uncertain to human review.

Ops discipline

Secrets as refs, tenant isolation, fail-loud error handling, rollback plans. Habits from running systems at scale.

AI where it earns it

Claude for extraction, insight, and agent tooling, with deterministic validation around the model, not blind trust.

Repositories are private. Live links, architecture, and outcomes are shared openly; source and infrastructure detail available on request under NDA.