AI / ML & LLM Engineering
Evidence-grade retrieval and agentic systems — hybrid search, claim-level verification, and LLM platforms built to prove what they claim instead of guessing.
RAG / GraphRAGLangGraphvLLMPyTorchMCP
About / The person behind the work
Dayan Ibrar.
AI & machine learning research engineer
I build the software, data infrastructure, and AI that turn complex information into useful decisions. My work spans federal and aerospace programs, financial systems, and independent research into reliable machine learning.
A full-stack engineer by practice, a researcher in how I approach a problem. I follow it from the underlying mathematics and data through to the application someone depends on.
Background / From curiosity to research
Building since twelve.
I started coding at twelve, building apps and publishing games. I still get the same satisfaction from turning an idea into something people can use. Today, I bring that drive to full-stack software, data engineering, and AI and machine learning. The problems have grown: aircraft readiness, financial data, complex enterprise systems, and the evidence behind an AI-generated answer.
My professional work spans defense, aerospace, and federal programs. At SteerBridge, I engineered data infrastructure supporting the USMC F-35 program, then built core data infrastructure for a Veterans Affairs claims program as a Senior ML/Data Engineer. My experience also includes contract work on proposals, prototypes, and architecture for Lockheed Martin and Sikorsky, and AI solution development through the IBM Partner Program. At Accenture Federal Services, I now build agentic and financial systems, bringing data engineering and applied AI into workflows where security, traceability, and sound decisions matter.
I work across the whole system: the interface someone uses, the APIs and services behind it, the pipelines that prepare its data, the models it depends on, and the infrastructure that keeps it running. Python, Rust, Go, and TypeScript are part of that work, alongside distributed processing, feature engineering, retrieval, model evaluation, and secure AWS architecture. Being able to follow a problem across those layers shapes how I design a solution.
My research starts with a question I can test. What does the data actually support? Does a neural model improve on a simpler baseline? Can an answer be traced to its sources? I use mathematical reasoning, statistical evaluation, and reproducible experiments to examine those questions. That direction connects my engineering experience to a growing interest in biomedical data and research tools, where the quality of the evidence determines what we can responsibly conclude.
Independent projects give me room to pursue those ideas in depth. I'm developing Structura around source-addressable retrieval, Praxis around governed AI-assisted development, and AegisOS around computing when connectivity is limited. Guardis, Relay, SignalForge, and ViDaT extend that work into security, data integration, signal processing, and visualization. I write about the decisions and limitations as the work develops.
I'm pursuing a Bachelor of Arts in Computer Science at Virginia Commonwealth University, expected in 2026, and a Master of Science in Computer Science at the University of Colorado Boulder, expected in 2027. My credentials span data engineering, machine learning, cloud architecture, and generative AI. I'm building on that foundation through research and hands-on engineering, with a consistent aim: make difficult problems understandable, then build systems people can use, inspect, and improve.
What I do
Four disciplines, one standard.
I work where AI, data, and secure infrastructure meet — on systems that have to be right the first time, for people who can't afford for them not to be.
Data Engineering at Scale
Multi-stage pipelines and lakehouse infrastructure moving petabytes — streaming ingestion, transforms, and provenance for systems where correctness isn't optional.
SparkKafkaAirflowdbtSageMaker
Secure Cloud & GovCloud
Production infrastructure on AWS GovCloud — FedRAMP-aligned, zero-trust, IaC-driven systems supporting defense and government programs.
AWS GovCloudTerraformLambdaFedRAMP
Systems & Platforms
Rust and Python at the core — my flagship systems Structura, Praxis, and AegisOS, plus full-stack products delivered end to end.
RustPythonGogRPC
Delivered across
Accenture Federal ServicesLockheed MartinSikorskyIBM Partner ProgramSteerBridge
Skills
Technical proficiency.
A breakdown of the tools and technologies I work with daily, rated by depth of production experience.
◆
Languages
Python95%
SQL92%
TypeScript85%
Rust78%
Scala75%
Go72%
Avg Proficiency83%
■
Data Engineering
Apache Spark92%
Apache Kafka88%
ETL Pipelines94%
Redshift85%
PostgreSQL90%
Redpanda78%
Avg Proficiency88%
▲
Cloud & AWS
AWS Lambda94%
AWS GovCloud90%
SageMaker85%
AWS Glue88%
S3 / SQS / SNS92%
Terraform87%
Avg Proficiency89%
▼
DevOps & Infra
Docker90%
Kubernetes82%
GitLab CI/CD88%
IaC / Terraform87%
FedRAMP Standards80%
mTLS / WSSE78%
Avg Proficiency84%
⬡
AI & ML
LangChain / RAG88%
PyTorch82%
IBM Watson85%
OpenAI API87%
TensorRT76%
Computer Vision78%
Avg Proficiency83%
Career
Experience.
Generative AI Engineer | Full-Stack LLM Development Specialist
Accenture Federal ServicesFederal fraud analytics, production data pipelines, and research into agent-based AI solutions.
- Developing a federal fraud detection approach combining supervised models (XGBoost and logistic regression), anomaly detection, graph analysis, and rules; designing out-of-fold evaluation and controls against training and temporal data leakage.
- Engineering Salesforce-to-AWS pipelines for incremental ingestion, deduplication, and analytics in S3 and Redshift, with watermark-based updates, reconciliation, and orchestration for large case datasets.
- Supporting data lake operations through IAM and S3 policy work, AWS Glue pipeline optimization, and Athena analytics and connectors.
- Assessing AI capabilities and researching agent-based solutions for federal use cases; translating technical findings into implementation options for program leadership.
XGBoostLogistic regressionGraph analysisSalesforceS3RedshiftAWS GlueAthenaIAM
Senior ML/Data Engineer
SteerBridgeData infrastructure and ML integration for a U.S. Department of Veterans Affairs claims program.
- Built core data infrastructure from schema and ingestion through production integration. The resulting workflow reduced a claims-processing step from more than six hours to under 15 minutes.
- Designed SOAP and REST integrations and pipelines for veteran and claims data; explored NLP, retrieval, and ML applications for document and claim workflows.
- Designed PostgreSQL schemas and modernized legacy SOAP service integrations for more reliable application and analytics access.
- Integrated AWS Glue and SageMaker components for claims and disability benefits questionnaire workflows; collaborated with engineers, data scientists, and VA stakeholders on architecture and compliance requirements.
PostgreSQLSOAP / RESTAWS GlueSageMakerNLPData modelingETL / ELT
F-35 Data Engineer
SteerBridgeSustainment data engineering and operational analytics for the F-35 program.
- Engineered pipelines for maintenance, telemetry, and readiness data used in operational analysis.
- Developed QuickSight dashboards and connected Azure Databricks and Data Factory workflows for defense data analysis.
QuickSightAzure DatabricksAzure Data FactoryData pipelinesOperational analytics
Lead Data Engineer
FinoSync CorpPrototypes and architecture for legacy-service modernization and aviation data platforms.
- Prototyped Relay, a gateway for exposing legacy SOAP and WSDL services through modern APIs, with authentication, permissions, and request audit concepts.
- Designed ViDaT Gov, a proposed aviation data platform covering ETL automation, analytics, and synthetic data workflows on AWS GovCloud.
SOAP / WSDLAPIsAWS GovCloudETLData architecture
Data Engineer (Contract)
Lockheed MartinProofs of concept and architecture proposals developed through Lockheed Martin supplier channels.
- Developed concepts for aviation sustainment, radar anomaly detection, and cyber applications.
- Submitted SBIR proposals, including AF254-0812, and developed concepts for signal analysis, supply chain optimization, and threat modeling.
Signal analysisAnomaly detectionArchitectureThreat modeling
AI Software Engineer (IBM Partner) · Software Engineer (AWS Partner)
IBM / Amazon Web ServicesPartner-program work across enterprise, government, aviation, and defense use cases.
- Developed IBM Watson Assistant and Discovery integrations and NLP automation for enterprise and government use cases.
- Worked with AWS partner teams on GovCloud, Lambda, S3, and SageMaker architecture for aviation and defense use cases.
IBM Watson AssistantIBM Watson DiscoveryNLPAWS GovCloudLambdaS3SageMaker
Lead Software Engineer
Patterns Tech LLCLed full-stack engineering and cloud infrastructure across a growing software consultancy, delivering scalable MERN and data engineering solutions.
- Built and deployed full-stack MERN applications improving user engagement
- Delivered real-time analytics dashboards for executive leadership
- Architected scalable microservices reducing infrastructure overhead
- Integrated AWS Glue, Athena, and cloud data warehousing workflows
MERN StackAWSAWS GlueAthenaData EngineeringCI/CD
Director of Technology (CTO)
StagewayServed as CTO of a non-profit organization, leading all technology and software development operations and defining long-term technology strategy.
- Led transition to Agile methodologies improving delivery consistency
- Implemented DevOps practices streamlining deployment pipelines
- Defined and executed a long-term technology roadmap
- Managed cross-functional engineering teams and stakeholder alignment
AWSDevOpsAgileFull-StackIT Program Management
Software Engineer
elo PitchConducted research into emerging technologies and built practical software prototypes across the full development lifecycle.
- Researched emerging technologies and translated findings into working software
- Built cross-platform prototypes with React Native and React.js
- Contributed across full lifecycle from requirements through deployment
- Delivered quality software across development, testing, and release
React.jsReact NativePythonPHPStripeData Engineering
Founder
Quad GrafixFounded a 3D modeling and open source printing organization, building software solutions that connected with Raspberry Pi to enhance the 3D printing process.
- Built 3D modeling solutions and open source 3D printing software
- Integrated Raspberry Pi for enhanced printing process control
- Published open source tooling for the maker community
- Combined hardware and software engineering from the ground up
UnityAWSRaspberry Pi3D ModelingOpen Source
Credentials & collaborations
Cleared, credentialed, trusted.
The work happens where precision isn't optional — defense, aerospace, and government programs, delivered under real accountability.
- Clearance
- Secret
- active
- Graduate education
- University of Colorado Boulder
- MS in Computer Science · In progress · Expected 2027
- Undergraduate education
- Virginia Commonwealth University
- BA in Computer Science · In progress · Expected 2026
- Based
- United States
- remote-ready
Accenture Federal ServicesAgentic and financial systems
USMC F-35 ProgramData engineering for fleet readiness & sustainment
Lockheed MartinDefense contract delivery
SikorskyDefense contract delivery
IBM Partner ProgramAI solution delivery
SteerBridgeSecure data infrastructure for mission programs
Current work / Engineering & investigation
The questions
I'm working on.
Reliable retrieval, accountable agents, and useful AI beyond the cloud. These independent systems give me room to work through each problem in depth.
In professional practice
Accenture Federal Services
Agentic and financial systems for federal programs, connecting data, applications, and secure infrastructure.
01 / Retrieval & evidence
Structura
Can an answer lead back to its source?
Developing retrieval that keeps source locations and evidence accessible through the path to an answer.
Read the system notes
02 / AI-assisted development
Praxis
What should an agent be allowed to change?
Building development workflows around explicit context, permissions, evaluation, and reviewable changes.
Read the system notes
03 / Computing at the edge
AegisOS
What remains useful without a network?
Exploring local AI, retained evidence, and recovery when connectivity is limited or interrupted.
Read the system notes
Philosophy / How I make decisions
Principles that
show up in the work.
The discipline I bring to an aircraft data pipeline also belongs in a web application or an ML experiment: clear assumptions, deliberate interfaces, and evidence that the result holds.
Follow the whole system.
A choice in the interface can change an API contract, a query plan, or an operator’s workload. I trace those relationships before optimizing one component.
Make the request path, ownership, and dependencies visible.
Start with the evidence.
For ML and AI, I want to understand the objective, the available information, and the baseline. A useful result survives inspection of the examples where it fails.
State the assumptions. Compare fairly. Keep the error analysis.
Design for recovery.
Data arrives late, requests repeat, and services become unavailable. I treat retries, observability, and rollback as design decisions that deserve their own tests.
Explain how the system detects a failure and returns to a known state.
Leave the reasoning behind.
I document the tradeoffs and revise them as new evidence arrives. Writing, open-source work, and focused experiments help turn an individual lesson into something another engineer can use.
Keep a decision record that someone else can challenge and improve.
Experience / Defense & aerospace
Engineering with
operational consequences.
Supporting aerospace programs has shaped how I think about software: understand the people using the information, preserve its context, and make the system possible to inspect and maintain.
Aircraft study / F-35Illustrative wireframe
Readiness begins with the data.
My work supporting the USMC F-35 program at SteerBridge connects data pipelines and analytics to fleet sustainment. The source, meaning, and freshness of a record matter to the decisions built on it.
Security is an architectural concern.
I consider identity, access boundaries, and data movement alongside the application itself. An integration needs a clear account of who can use it and what it exposes.
Traceability makes review possible.
Documented interfaces, lineage, and explicit failure handling help another engineer understand what changed, why it changed, and how to investigate a problem.
Method / From question to operation
Make the reasoning
part of the build.
I want the finished system to carry an explanation of how it works, what it assumes, and how it was tested. That applies to a frontend, a data platform, and a trained model.
The engine study is a reminder to look at the interfaces between parts. In software, that means examining contracts, dependencies, and what happens when one stage behaves differently than expected.
Mechanical study / TurbofanStatic drawing
Frame the problem.
Identify the user, the decision, and the operating constraints. Examine the existing data and define what a useful result would look like.
Problem statement + acceptance criteriaMake the design explicit.
Map interfaces, data flow, and trust boundaries. For ML, define the baseline, evaluation split, and assumptions before comparing approaches.
Architecture + evaluation planBuild and challenge it.
Implement in reviewable increments. Exercise edge cases, inspect failures, and measure the behavior that matters to the original question.
Working implementation + test evidenceOperate and learn.
Deploy with a recovery path. Monitor real behavior, investigate unexpected results, and bring those observations into the next design decision.
Runbook + observations + revisions
Building innovative systems
GitHub activity.
A snapshot of my contributions and coding activity over the past year — building tools, shipping features, for private orgs, research & Development and more.
0+
Total Commits
Across private & public repos including FinoSync Corp, AWS, and defense projects
0+
Active Days
Days with contributions since 2019
0
Longest Streak
Consecutive days of shipping code
Interests
Beyond the code.
The things that fuel my engineering work and keep me curious — from writing and open source to aviation and AI research.
Technical Writing
I publish deep-dives on architecture, AI pipelines, and engineering best practices. Writing clarifies thinking and builds community.
Open Source
Contributing to the tools I rely on and building new ones. OSS is how I give back to the community that taught me to code.
Aviation & Aerospace
A lifelong fascination with flight, from aerodynamics to avionics. It shapes how I think about precision, systems, and reliability.
AI & Agents
Experimenting with autonomous agents, RAG architectures, and the frontier of what software can do when it can reason.
Mentoring
Helping early-career engineers navigate the industry through code reviews, pair programming, and knowledge sharing.
Continuous Learning
Currently exploring Rust systems programming, edge computing, and ML infrastructure. The learning never stops.
Let's build
Systems that matter, built to spec.
Evidence-grade AI, mission-critical infrastructure, or a platform that needs to be right the first time — if it has to hold weight, let's talk.