Complexity Labs, Inc.

What NRV does

Demonstrated on public federal data

Each demonstration runs end to end on request, live, on a question the viewer chooses. All use public data only; no agency system was accessed.

• Motor carrier identity: five Federal Motor Carrier Safety Administration datasets (out of service orders, the Company Census, roadside inspections, the crash file, and Licensing and Insurance) joined to link carriers registered around or after a serious out of service order to the carrier ordered off the road, graded by how many independent signals they share, with the roadside record since. Every link is a reason to look, never a finding.
• Across operating administrations: the Federal Highway Administration, Federal Motor Carrier Safety Administration, Federal Railroad Administration and Federal Transit Administration each write a state differently. Reconciled once, they produce motor carrier crashes per billion vehicle miles travelled by state, a figure none of them produces alone.
• Across agencies, with no shared key: motor carrier crash and inspection records joined to National Highway Traffic Safety Administration vehicle identification, recall and investigation records, by what a make or a failed part is called in each system.
• Four NASA archives: the Common Metadata Repository, the Technical Reports Server, the Open Science Data Repository and TechPort, four schemas with no common identifier, joined into one queryable view with lineage on every field.
• Four kinds of system at once: a PostgreSQL database, object storage, a live Kafka stream and a REST service, federated into one graph and served through the read-only query service.

Working software

NRV is running today in closed beta. 240 automated tests pass on every run, backed by integration trials against real PostgreSQL, S3-compatible object storage and Kafka-compatible streaming brokers. It is built to slot in behind an agency’s own gateway and identity provider, so access control stays with the systems you already trust.

Differentiators

• We extend, we do not replace. The systems you run today keep serving traffic. NRV reaches across them, so no office has to give up a system or a budget line for a question to be answered.
• Built for your tenant. NRV is designed to run in your cloud account, on your storage, behind your identity provider and gateway.
• Integration through published interfaces. We connect to source systems the way they already expect to be read, and leave them unmodified.
• The judgement stays with your staff. Thresholds, crosswalks and matching rules live in small, readable files that your people review and change, not inside a model.
• No lock-in. Plain SQL, Python and open file formats throughout. You can take the work in-house, or take it elsewhere.
• Aligned to NIST. The Cybersecurity Framework, Special Publication 800-53 controls and the AI Risk Management Framework, treated as design inputs rather than a compliance pass at the end.

Engineering around NRV

The services we deliver so that NRV can be put to work on an agency’s own data.

• Data modernization: moving data off aging systems into the cloud, every record verified, none lost, and federal retention rules preserved.
• Data platform engineering: pipelines, layered data stores and live feeds in place of overnight reports, on Amazon Web Services and Microsoft Azure.
• Machine learning: from raw data to models in production, watched for drift, with data-quality checks built into the pipeline.
• Post-quantum readiness: inventory and migration planning for the new federal standards, ML-KEM (FIPS 203) and ML-DSA (FIPS 204).

Delivery record

Delivered by our engineering leadership on regulated financial platforms at prior employers, at institutional scale. Stated qualitatively: these were their engagements, not Complexity Labs contracts.

• Daily investment accounting: built and ran the cloud data pipeline, roughly 100 million records a day, behind daily investment accounting at a global asset manager, feeding valuation checks, settlement reconciliation and daily profit and loss.
• Records migration: moved 50 terabytes of records off on-premise storage into the Microsoft and Amazon clouds, every record verified and federal retention rules (Securities and Exchange Commission Rule 17a-4, FINRA) preserved.
• Regulatory reporting: built the engine that slices portfolios for capital and solvency reporting, reconciling several independent ratings and classification systems into one view.

Who you work with

• Research training: graduate training in computer science at Columbia University; doctoral research in the theory of machine learning at the University of Cambridge, beginning in Lent term 2027; and peer-reviewed work published by the Institute of Electrical and Electronics Engineers on traffic intersection video analytics.
• Certified, clearance eligible: AWS Certified Data Engineer and Cloud Practitioner. United States citizens, clearance eligible.

How we deliver

We start from NRV rather than from a blank repository, so an engagement is spent on your data and your rules instead of on plumbing. Work lands in your environment continuously, in small reviewable pieces, against your security review rather than around it. What we hand over is running software, its lineage, its tests and its documentation, on interfaces your own engineers can pick up.

Registration and contact

UEI

SACZBXUXLAL5

CAGE

16YX0

NAICS

541511 (primary) · 541512 · 541519

Entity

Delaware corporation, incorporated 2025

Status

Small business and Small Disadvantaged Business, self-certified in SAM.gov · SAM.gov active

Contact

Ashvin Jagadeesan, President

Email

ashvin@complexitylabs.co

Web

complexitylabs.co

Technology

NRV: Python · SQL · DuckDB · Apache Arrow · Parquet · PostgreSQL · Apache Kafka · S3-compatible object storage · Docker. Delivery: Amazon Web Services · Microsoft Azure · Microsoft Fabric · PySpark · Delta Lake · Snowflake · PyTorch · scikit-learn