SV
Open to AI Engineering, Machine Learning, and Generative AI opportunities — remote or on-site, and willing to relocate.

Building Intelligent Systems That Turn Data Into Action

Mid-Level AI Engineer specializing in AI-powered applications, generative AI, machine learning, intelligent automation, data systems, and scalable production solutions.

AI Engineer · Boone, NC — open to remote & willing to relocate

Explore My Work
About

AI engineering that ends
in production, not a demo.

I turn ambiguous business problems into working AI systems — then keep them reliable once real users show up.

I'm an AI Engineer with a Master's in Computer Information Technology from Purdue University. My work spans LLM integration and RAG pipelines, multi-agent orchestration, data platforms, and full-stack delivery on AWS.

I've built AI concierges, automated marketing pipelines, data warehouses, and real-time mobile platforms — embedded directly with the stakeholders who use them, all in production, all measured by outcomes.

Translate business problems into AI solutions
Build AI-powered applications end to end
Work with structured and unstructured data
Develop generative AI and machine-learning workflows
Integrate APIs, models, databases, and cloud services
Ship scalable, maintainable, production-ready systems
Communicate technical findings to stakeholders
0+
Months production AI experience
0
AI & data projects shipped end to end
0+
Technologies used in production
0+
Data sources unified & automated
Portrait of Saiabhiram Velamakanni, AI Engineer
Saiabhiram Velamakanni
M.S. Computer Information Technology · Purdue University · May 2024
Distributed systems, data engineering, machine learning, and cloud architecture.
Technical Skills

A connected stack, not
a list of buzzwords.

Five layers that work together — from model behaviour down to the infrastructure it runs on.

Artificial Intelligence

Generative AI systems that stay grounded in real, current context.

  • Generative AI01
  • Large Language Models02
  • Prompt Engineering03
  • Retrieval-Augmented Generation04
  • AI Agents (LangGraph)05
  • Natural Language Processing06
  • Model Evaluation (RAGAS)07
  • Function Calling / Tool Use08
  • Streaming Responses09
  • Computer Vision10
Experience

Embedded AI engineering,
in real client environments.

Outcomes over duties — every engagement ends with a system running in production.

AI Engineer

Oct 2024 — Present
Blue Ridge Preservation · Boone, NC · Full-Time

Embedded AI engineering partner — delivering end-to-end AI systems directly in client environments across data infrastructure, agentic AI, and full-stack product development.

  • Designed and shipped 4 complete AI transformation projects from discovery through production
  • Built production LLM systems with Claude — context injection, RAG, function calling, multi-agent orchestration
  • Architected AWS infrastructure (EC2, RDS, Lambda, CloudFront, ECS, SNS) with Terraform and CI/CD
  • Automated manual marketing and reporting workflows with a multi-agent pipeline
  • Unified 6+ fragmented data sources into a governed warehouse with automated quality checks
  • Delivered real-time systems on Firebase RTDB for sub-5-second sync across client surfaces
  • Collaborated directly with non-technical stakeholders and translated findings into decisions
Claude APILangGraphPythonFastAPIAWSTerraformPostgreSQLBigQuerydbtReact Native
Selected Work

Systems in production.
Not proofs of concept.

Each project below shipped end to end — problem framing, AI architecture, deployment, and the outcome it moved.

Project 01

AI-Powered Guest Experience Platform

One mobile platform for guests and staff, with an LLM concierge and live shuttle tracking.

Generative AIFull-Stack AICloud
Problem

A hospitality campus ran day visitors, registered guests, and operations staff through disconnected tools, phone calls, and paper requests — with no live view of shuttles or open tasks.

AI solution

A mobile-first platform with dual-mode RBAC: an AI concierge (Aranya) built on Claude with live context injection, an intelligent request router, and zero-hardware GPS tracking from driver phones.

Architecture

React Native client → Node.js API on AWS → PostgreSQL (RDS) for bookings and CRM, Firebase RTDB for real-time GPS, Claude API for the concierge with context assembled per request.

  • AI concierge answering guest questions from always-current campus context — no fine-tuning
  • Real-time shuttle tracking: driver phones broadcast GPS to Firebase RTDB every 5 seconds
  • Request router classifying guest requests into Operations and Facilities dashboards
  • OTP auth verified against the live booking registry — no passwords, no account creation
Business impact
Sub-5s GPS updates
$0 GPS hardware cost
1,000+ visitors in CRM
React NativeClaude APIFirebase RTDBAWSPostgreSQLNode.jsSendGrid
GitHubLive demo — private client system
Project 02

Agentic AI Marketing Automation

A 7-agent LangGraph system replacing a manual marketing and attribution workflow.

AI AutomationGenerative AIData Analytics
Problem

Content creation, campaign management, and attribution were manual and fragmented — with no single view of return on ad spend across channels.

AI solution

A supervisor agent routes work to six specialist agents with conditional handoffs based on completion state, normalising everything through a central PostgreSQL schema.

Architecture

LangGraph supervisor graph → 6 specialist agents → FastAPI service layer → PostgreSQL schema → HubSpot and channel APIs, with human approval gates before publishing.

  • Supervisor graph with conditional routing — not a hardcoded sequence
  • First single-view ROAS attribution across all marketing channels
  • Human-in-the-loop checkpoints before any external publishing
  • Normalised data flow through a central PostgreSQL schema
Business impact
~20 hrs/week saved
Unified ROAS view
Human-in-the-loop by default
LangGraphClaude APIPythonFastAPIPostgreSQLHubSpot API
GitHubLive demo — private client system
Project 03

Unified Data Warehouse & AI Analytics

Fragmented operational sources turned into one warehouse with a natural-language query layer.

Data AnalyticsCloudGenerative AI
Problem

Bookings, marketing, POS, and spa operations lived in separate systems. Answering a basic revenue question meant manual exports and reconciliation.

AI solution

A modern ELT warehouse with a dbt semantic layer and a Claude-powered NL→SQL interface so non-technical staff can ask questions directly.

Architecture

Fivetran/Airbyte → BigQuery → dbt models and tests → Looker dashboards, with a Claude NL→SQL layer querying the governed semantic models. Orchestrated with Airflow.

  • ELT pipeline: Fivetran/Airbyte → BigQuery → dbt → Looker
  • Semantic layer modelling guests, bookings, revenue, and marketing spend
  • Claude-powered NL→SQL query interface for non-technical staff
  • Data quality enforced with dbt tests and Great Expectations
Business impact
6 sources unified
Org-wide ROAS & LTV
Natural-language query layer
BigQuerydbtFivetranAirbyteLookerClaude APIAirflow
GitHubLive demo — private client system
AI Lab

Ask My Portfolio.
A live retrieval demo.

A grounded assistant that answers questions about my experience, projects, and stack using only this portfolio's content.

Portfolio Assistantgrounded · gemini
Ask me anything about Saiabhiram's experience, projects, or stack — I answer from his portfolio content.
Engineering Approach

How I take an AI idea
from problem to production.

A repeatable path. It keeps scope honest, cost visible, and reliability measurable.

01Step 1

Understand the business problem

Start with the decision or workflow that needs to change — not the model.

02Step 2

Evaluate the available data

Assess coverage, quality, and freshness before promising an outcome.

03Step 3

Select the appropriate AI approach

Rules, classical ML, retrieval, or agents — whichever is the smallest thing that works.

04Step 4

Build and test the solution

Ship a thin vertical slice, then harden it against real inputs.

05Step 5

Evaluate accuracy, cost, latency, reliability

Every AI feature gets measured on all four, not just quality.

06Step 6

Deploy the system

Containerised, provisioned as code, released through CI/CD.

07Step 7

Monitor and improve

Instrument outputs, watch for drift, and iterate on prompts and models.

For Hiring Teams

What I bring
to an AI team.

Open to AI Engineering, Machine Learning, and Generative AI opportunities — remote or on-site, and willing to relocate.

Practical AI implementation

Four production systems shipped from discovery through deployment — not notebooks.

Strong Python and SQL foundation

Comfortable across pandas, FastAPI, warehouse modelling, and window-function SQL.

Models connected to real applications

LLM features wired into mobile apps, dashboards, and internal operational tooling.

Full-surface experience

Data pipelines, APIs, cloud infrastructure, and user interfaces in the same project.

Production reliability mindset

Guardrails, evals, monitoring, and human-in-the-loop where correctness matters.

Stakeholder communication

Embedded directly with non-technical clients and translated findings into decisions.

Continuous learning

Actively deepening AWS, data engineering, and advanced LLM patterns.

Contact me
Contact

Let's build something
that ships.

Open to AI engineering roles, consulting engagements, and forward-deployed positions. If you need production AI systems — not demos — let's talk.

saiabhiram369@gmail.com LinkedIn GitHub
Boone, NC — open to remote & willing to relocate