AI Engineer / Builder / Founder

Building intelligent systemsthat actually ship.

AI agents, automation systems, full-stack products, and verifiable Web3 infrastructure.

View ProjectsGitHub
KERALA, INDIAAVAILABLE FOR INTERESTING BUILDS
SYSTEM MAP
AKILAI AGENTSNEBULASTELLARZK SYSTEMSFULL-STACKOBSILOOM
HOVER A NODE TO INSPECT
What I Build

Systems, not demos.

Six areas that most of my work falls into — from language model applications and autonomous agents through to verifiable on-chain infrastructure.

Akil Ayman — AI Engineer and Web3 Developer
  • 01

    AI Systems

    LLM applications with retrieval, evaluation and grounding — built to survive contact with real inputs rather than demo prompts.

    LLMRAGFAISSEVALUATION
  • 02

    AI Agents

    Autonomous agents with tool access, MCP integrations and decision loops that know when to act and when to hand back.

    AGENTSMCPTOOL USE
  • 03

    Automation

    Business process automation that removes repeated human steps from sales, operations and support workflows end to end.

    N8NZAPIERWORKFLOWS
  • 04

    Web3 Infrastructure

    Soroban smart contracts and Stellar-native systems — oracles, vaults, payment rails and inter-contract execution.

    STELLARSOROBANRUST
  • 05

    Zero-Knowledge Systems

    Verifiable computation with RISC Zero — proving a result is correct without revealing what produced it.

    RISC ZEROZKVMPROOFS
  • 06

    Full-Stack Products

    Complete products from schema to interface — typed APIs, real data models, and frontends that hold up under use.

    NEXT.JSFASTAPIPOSTGRESQL
Experience

Where the work happens

Building an AI automation venture, alongside applied AI engineering work across energy and agricultural operations.

  1. 012026 — PRESENTCURRENT

    ObsiLoom

    Founder & AI Automation Engineer

    Building an AI automation agency focused on custom AI agents, intelligent workflows, and business process automation.

    • Custom AI agents
    • Intelligent workflow automation
    • AI-powered business processes
    • Sales automation
    • Operations automation
    • Customer support automation
    LLM AGENTSMCPRAGN8NFASTAPIGCP
  2. 02JUNE 2026 — JULY 2026

    Hastus Energy

    AI Engineer

    Developed AI-driven automation workflows and intelligent systems for business and operational use cases.

    • AI-driven automation workflows for operational use cases
    • Integrated AI services and APIs to streamline workflows
    • Supported automated business processes with intelligent systems
    PYTHONAI APISWORKFLOW AUTOMATION
  3. 03JUNE 2026 — JULY 2026

    Shri Raja Ramtek FPO Ltd.

    AI Intern

    Worked on practical AI and machine learning use cases involving agricultural and business operations.

    • Practical AI and machine learning use cases
    • Data processing and AI-based workflows
    • Operational decision-making support
    PYTHONMACHINE LEARNINGDATA PROCESSING
THE VENTURE

OBSILOOM

AI Automation for Businesses

Custom AI agents and intelligent workflows that take repeated business processes off people's desks — across sales, operations and customer support.

AGENT WORKFLOWLIVE

  1. TRIGGER

    event · webhook · schedule

  2. AI AGENT

    reasoning loop

  3. TOOLS

    APIs · data · systems

  4. DECISION

    route · escalate · act

  5. ACTION

    execute

  6. BUSINESS RESULT

    outcome

01

Lead Qualification

Inbound leads scored, enriched and routed automatically before anyone opens the inbox.

02

Sales Automation

Follow-ups, CRM hygiene and pipeline movement handled by agents instead of reminders.

03

Customer Support

Tiered response agents that resolve the routine and escalate the rest with full context.

04

Workflow Automation

Multi-step internal processes wired end to end across the tools a business already runs.

Featured Build2026

NEBULA

Verifiable Compute Layer for Stellar

Compute off-chain.
Verify on-chain.

A zero-knowledge verifiable compute layer for Stellar. Computationally intensive workloads are offloaded to an off-chain execution environment, and their results are cryptographically verified on-chain.

nebula_stellar
PIPELINE
  1. 01USER / APPLICATION

    REST client

    A client submits a compute job over the REST API.

  2. 02COMPUTE JOB

    queued · persisted

    The job is persisted to PostgreSQL and enters the state machine.

  3. 03OFF-CHAIN EXECUTION

    executor service

    The executor service runs the workload off-chain.

  4. 04RISC ZERO ZKVM

    guest program

    The guest program executes inside the RISC Zero zkVM.

  5. 05ZERO-KNOWLEDGE PROOF

    succinct receipt

    A succinct receipt is produced over the execution trace.

  6. 06SOROBAN VERIFICATION

    on-chain contract

    The proof digest is submitted to a Soroban verification contract.

  7. 07STELLAR

    settled

    Stellar confirms. Private inputs never left the off-chain environment.

The Problem

Smart contract platforms cap how much computation a transaction can do. Anything heavy — risk models, identity checks, cross-chain header validation — has to run off-chain, and once it does there is no way for the ledger to trust the result.

The Approach

Nebula runs the workload inside a RISC Zero zkVM, produces a succinct zero-knowledge proof of correct execution, and submits that proof to a Soroban contract for on-chain verification. Private inputs never leave the off-chain environment — only public commitments and proof digests reach the ledger. A Rust/Axum orchestrator drives jobs through a typed state machine with SSE streaming for real-time status.

Architecture

  1. 01REST API accepts a compute job and persists it to PostgreSQL via SQLx
  2. 02Executor runs the matching guest program inside the RISC Zero zkVM
  3. 03Proof service produces a succinct receipt over the execution trace
  4. 04Verification client submits the proof digest to a Soroban contract
  5. 05Stellar testnet confirms; job state streams back over SSE

Stack

RUST
01
AXUM
02
POSTGRESQL
03
SQLX
04
RISC ZERO
05
ZKVM
06
SOROBAN
07
STELLAR
08
ZERO-KNOWLEDGE PROOFS
09
SSE
10
JOB ORCHESTRATION
11
The Archive

Selected builds

Seven additional builds across AI, Web3 and full-stack product work. Every one is open source — the descriptions come from what the repositories actually contain.

Stack

The tools underneath

What I actually reach for. No percentages — these are either things I've shipped with or things I haven't.

01

AI & Machine Learning

  • LLM Application Development
  • RAG Pipeline Design
  • FAISS Vector Stores
  • Autonomous AI Agents
  • MCP Integrations
  • Fine-Tuning
  • QLoRA
  • Unsloth
  • Scoring & Ranking Algorithms
  • Prompt Engineering
02

Cloud & Infrastructure

  • GCP
  • Cloud Run
  • Vertex AI
  • Cloud Storage
  • Docker
  • Microservice Architecture
  • CI/CD
  • REST APIs
  • API Gateway Patterns
  • Linux / Bash
03

Full-Stack Development

  • Python
  • FastAPI
  • Flask
  • JavaScript
  • TypeScript
  • Next.js
  • React
  • PostgreSQL
  • Firebase
  • n8n
  • Zapier
04

Web3

  • Stellar
  • Soroban
  • Rust
  • Zero-Knowledge Proofs
  • RISC Zero
  • Midnight Network
Achievements

Recognition

WEB3 HACKATHON WINNER

Stellar BuildStation

KOZHIKODE

Won with Nebula — a zero-knowledge verifiable compute layer that offloads heavy workloads off-chain and settles their proofs on Stellar through Soroban.

See the case study
Certificates

Credentials

Verified IBM credentials. Every entry links to its live certificate page.

Education

Srinivas University

B.Tech. in Artificial Intelligence & Machine Learning

Let's buildsomething intelligent.

AI systems. Automation. Web3 infrastructure. Let's turn the idea into something real.

Email meGitHubLinkedIn
AKILAYMAAN@GMAIL.COM