Product thinking × mechanical engineering × software

Janani Manathattai

I’m a UCLA mechanical engineering student working across robotics, medical research instrumentation, and software.

Janani Manathattai standing under the UCLA Royce Hall arches
UCLA
B.S. Mechanical Engineering
Expected 2029
01 / WORK & PROJECTS

Problems I’ve framed, scoped, and built through.

Product proposals, intern-owned tools, robotics, and research instrumentation—organized around the decisions I made, the work I owned, and the evidence behind the result.

Healthcare software + data standardsVetsEZ / EZ Labs · 2026

CareClaims Intake Gateway

CareClaims combines healthcare-claim intake, standards validation, correction guidance, and downstream regulatory processing in one product. I owned the intake gateway for X12 EDI and CMS-1500 submissions, adapting an open-source X12 foundation to client and payer requirements while reducing the active contributors needed for the workstream from four to two.

  • Converted client needs into intake rules: met with the claims client to identify required fields, payer-specific 837 variations, acceptance criteria, and correction outputs, then translated those requirements into configurable validation behavior.
  • Adapted the open-source X12 framework: evaluated available parsers, selected the strongest standards-aware foundation, and extended its segment mapping and validation logic around TR3 guidance so claims from different health insurers normalized into the same client workflow.
  • Implemented the intake pipeline: built HTTP receipt, empty-payload rejection, EDI normalization, batch parsing, submission and claim-state persistence, configurable rule execution, and generation of X12 999 acknowledgments for 837 transactions.
  • Made errors actionable: mapped IK3, IK4, CTX, IK5, and AK9 acknowledgment data into summaries that identify the failing segment, element, position, code, and correction context for healthcare facilities.
  • Stabilized concurrent behavior: added automated tests for empty requests, rule toggling during active intake, acknowledgment defects, delimiter edge cases, and unknown dispositions so two contributors could maintain the workstream reliably.
Product ownershipOpen-source evaluationX12 EDICMS-1500TR3 guidanceError UX
4 → 2 peopleContributors needed after making the feature set more efficient and maintainable
One gatewayEDI and CMS-1500 intake paths feeding a unified regulatory workflow
837 → 999Standards-aware ingestion, validation, acknowledgment, and correction guidance
2 audiencesMachine-readable compliance output plus actionable facility-facing explanations
Full-stack product engineeringVetsEZ / EZ Labs · 2026

Subscription & License Management Platform

I owned an internal subscription and license platform that replaced scattered records with searchable seat-level inventory, cost controls, approvals, and audit history. By exposing frozen, unwanted, unassigned, and otherwise unaccounted-for licenses, the product helped reduce unnecessary company software expenditure by 10% and gave executives clearer evidence for renewal decisions.

  • Created the operating data model: designed PostgreSQL tables and REST APIs for vendors, products, subscriptions, individual seats, assignees, renewal dates, billing cycles, status changes, and immutable audit events.
  • Reduced avoidable spend by 10%: implemented unassigned-seat tracking, frozen and non-billable states, expiration sorting, employee-level search, and renewal visibility so administrators could remove or reallocate licenses that were unwanted or no longer attributable to an employee.
  • Built executive analytics: compared employee growth with license-expenditure growth, calculated cost per active seat and total product cost excluding frozen/non-billable licenses, ranked subscriptions by cost and utilization, and surfaced upcoming renewals to show which products delivered the most and least value.
  • Connected ticketing to fulfillment: linked DevOps requests to eligibility checks, available-seat counts, approve/decline decisions, create-and-assign flows, decision messages, seat assignment, and independent ticket closure.
  • Protected production workflows: added assignment row locking and 409 conflict handling, AES-256-GCM protection for PII, scheduled expiration notifications, integration tests, Docker services, and Azure deployment.
Product ownershipRequirementsReactTypeScriptPostgreSQLAzure
10% reductionDecrease in unnecessary license expenditure through seat-level accountability
Executive analyticsEmployee growth, license-spend growth, utilization, cost, value, and renewals in one view
6 core flowsSearch, assignment, reassignment, renewals, notifications, and auditing
Cloud-nativeReact and Node services backed by PostgreSQL and deployed through Azure
Mechanical design + roboticsSummer 2026
SolidWorks rendering of the Automated Quality Sorting Manipulator with conveyor, dual gantry, articulated arm, and gripper

Automated Quality Sorting Manipulator

In a four-person team, I developed the mechanical and electromechanical architecture for a modular system that sorts accepted and defective components from a conveyor. My work covered subassembly interfaces, actuation and power requirements, material and process selection, tolerance targets, and performance calculations used to determine the next prototype iteration.

  • Designed a 29-part assembly: integrated a belt conveyor, synchronized dual Y gantries, wheel-driven X translation, a 3-DOF articulated arm, and a motorized gripper into five independently testable subsystems in SolidWorks.
  • Specified actuation and power: mapped motor and servo loads, AC-to-DC conversion, voltage regulation, axis-level control, and synchronization requirements needed to prevent dual-gantry skew and linear-guide binding.
  • Defined manufacturable interfaces: selected 1060 aluminum for machined structural components, ABS for lower-mass printed parts, standard hardware where available, and ±0.002 inch tolerances at bearing, shaft, lead-screw, wheel, and pivot fits.
  • Calculated the motion constraint: modeled 1,681 N axial thrust from a 2.5 N·m actuator but only 22.7% lead-screw efficiency and 1.94 mm/s traverse speed, identifying cycle time—not force capacity—as the primary design limitation.
  • Specified physical validation: recommended a coarser or multi-start lead screw, followed by backlash, payload, positional accuracy, vibration, and coordinated-motion testing before integrating machine vision.
View the complete 86-page CAD and engineering report ↗
SolidWorks3-DOF actuationMotor + servo systemsDFMTolerance analysisPower architecture
29 parts / 5 systemsConveyor, Y gantry, X gantry, articulated arm, and gripper assembly
1,681 NPredicted axial thrust from a 2.5 N·m actuator under modeled conditions
±0.002 inCritical-interface tolerance target
22.7%Calculated lead-screw transmission efficiency
1.94 mm/sCalculated traverse speed and primary cycle-time constraint
31.43 kgModeled full-system mass across a 130.81 cm assembly length
Research-based competition proposalsTSA TEAMS · 2023 + 2025

TSA TEAMS Engineering Proposals

Two competition essays developed as researched engineering proposals rather than physical prototypes. Each response translated an open-ended national prompt into a defined location, technical system, quantified cost and performance assumptions, hazards, constraints, and an implementation or validation plan.

Proposal 01 · Urban transportation2023 · Nationals qualifier

Increasing Los Angeles Bicycle Adoption

2023 competition prompt: develop an engineering proposal to increase the population of bicycle users in a selected city. Our seven-person team proposed a theoretical Los Angeles bike-share and safety system—not a built prototype—that placed third in California and qualified for nationals.

  • Converted the prompt into requirements: evaluated Metro Bike adoption barriers across station density, cash access, checkout reliability, vehicle sizing, theft exposure, tire failures, route safety, and perceived convenience.
  • Specified the service architecture: proposed QR unlocking, destination and docking guidance, mobile and cash-kiosk payments, two frame sizes, storage options, mileage incentives, and recyclable airless tire inserts.
  • Analyzed hazards and failure modes: addressed flat tires, heat exposure, theft, unsafe intersections, vehicle conflicts, insufficient lane separation, and the operational consequences of unavailable docks.
  • Modeled implementation cost: scoped 100 stations with 25 bicycles each, ten miles of protected high-traffic corridor, associated station/app/infrastructure costs, and a total estimated program cost of $14 million.
  • Defined a validation plan: proposed a two-month ridership pilot with city data collection before permanent lane conversion, allowing adoption and safety results to determine whether the infrastructure should scale.
Read the full 2023 proposal ↗
Requirements analysisHazard analysisCost modelingPilot design
Proposal 02 · Climate-resilient infrastructure2025 · Nationals qualifier

Speyside Resort: Sustainable Caribbean Design

2025 competition prompt: design a sustainable resort in a tropical coastal location that could withstand extreme weather while using locally abundant natural resources. Our four-person team proposed a six-story resort in Speyside, Tobago that combines structural, environmental, and electrical systems to resist flooding and tropical storms while converting rainfall and energetic ocean waves into usable resources.

  • Researched wave-energy generation: evaluated SWEL’s Waveline Magnet, which converts vertical wave motion into electrical current through oscillating magnets and conductive coils; proposed two shoreline units rated at up to 100 kW each under optimal conditions to take advantage of the higher-amplitude waves produced by tropical storms.
  • Designed a second rainfall-energy path: specified dual-electrode D-TENG panels across a 50,000-square-foot, eight-sided roof so flowing rain droplets transfer charge through triboelectric and electrostatic induction, with the essay’s model estimating 25.39 kWh of annual generation.
  • Engineered the storm-resistant envelope: selected 12-gauge steel framing, fiber-cement cladding, fiberglass windows, fiberglass insulation, and phase-change materials; the modeled wall assembly increased R-value from 0.65 to 5.65 m²·K/W and reduced heat transfer by approximately 89%.
  • Integrated flood detection and response: used LiDAR-derived elevation maps and digital surface models to locate retention ponds and drains, then proposed 80 water-level sensors, 20 AI-enabled cameras, automated diversion controls, and self-stabilizing watergate barriers.
  • Closed the stormwater loop: routed collected water through retention and storage systems and proposed automated bioremediation using targeted bacterial cultures to break down hydrocarbons, heavy metals, and plastics before reuse.
  • Modeled feasibility and limitations: estimated $2.17 million for the steel structure and watergate system, $427,000 for LiDAR/AI integration, and $2.5–$3.75 million for D-TENG and two Waveline Magnets while identifying permitting, transport, maintenance, grid integration, and funding as implementation risks.
Read the full 2025 proposal ↗
Wave energyD-TENGLiDAR + AIFlood controlBioremediationCost analysis
2 qualifiersBoth researched proposals advanced the team to nationals
3rd in California2023 bicycle-adoption essay placement
$14M planModeled Los Angeles bicycle-system implementation
200 kWCombined optimal rating of two proposed Waveline Magnet units
89% reductionModeled decrease in heat transfer through the resort wall assembly
$2.5–$3.75MEstimated D-TENG and wave-energy implementation range
Medical instrumentation + computationUCLA Health / 2023–2026

Respiratory DAQ & 5DCT Validation

I modernized a respiratory data-acquisition and analysis workflow used by radiation-oncology researchers to model patient breathing patterns and validate 5DCT data. The LabVIEW changes enabled real-time monitoring and recording, expanded the data available for abnormality review, and improved practitioner breathing-model accuracy by 50%.

  • Rebuilt the acquisition chain: connected P409X respiratory transducers through DAQmx and modular SubVIs for calibration, voltage-to-pressure conversion, signal conditioning, digital filtering, waveform processing, and timestamped recording.
  • Improved modeling accuracy by 50%: added continuous real-time graphing and recording so practitioners could compare higher-resolution breathing curves, identify abnormal waveform behavior, and fit patient respiratory patterns more accurately.
  • Modernized legacy LabVIEW code: replaced outdated blocks with current open-source VIs, separated acquisition and visualization logic, optimized dataflow, and expanded graph inputs beyond the smaller dataset handled by the previous program.
  • Automated 5DCT validation in MATLAB: wrote functions to sort patient datasets, increase artifact visibility, flag likely registration errors, and separate false positives for researcher review.
  • Improved maintainability: parameterized Front Panel controls and documented SubVI inputs, outputs, units, dependencies, and expected signal ranges so later lab members could recalibrate or extend the workflow.
LabVIEW DAQmxSubVI architectureMATLABP409X transducersDigital filtering5DCT imaging
50% more accuratePractitioner modeling of patient breathing patterns after the LabVIEW upgrade
3+ yearsProgression from research intern to lab assistant
Real-time DAQCalibration, conversion, filtering, graphing, and timestamped recording
5DCT + MATLABDataset sorting, artifact inspection, registration-error detection, and false-positive review
02 / EXPERIENCE

Experience across software, research, and engineering.

Internship and research roles spanning healthcare software, instrumentation, data analysis, and aerospace maintenance.

Jun 2026 — Present

VetsEZ / EZ Labs

AI Product Engineering Intern · Remote
  • Owned two intern-led products across React, TypeScript, Node.js/Fastify, PostgreSQL, Docker, and Azure, translating stakeholder workflows into schemas, REST APIs, validation behavior, test plans, and release scope.
  • Met with the claims client to capture payer-specific 837 requirements, then adapted an open-source X12 framework and TR3-based rules for EDI/CMS-1500 intake, claim-state tracking, 999 acknowledgments, and facility-facing error summaries—reducing the CareClaims workstream from four contributors to two.
  • Reduced unnecessary software-license expenditure by 10% by surfacing frozen, unwanted, unassigned, expired, and employee-unaccounted seats for cancellation or reassignment.
  • Built an executive analytics page comparing employee growth with license-spend growth, product utilization, cost per active seat, total billable cost, value ranking, and upcoming renewals to support subscription purchasing and renewal decisions.
May 2023 — Jun 2026

UCLA David Geffen School of Medicine

Lab Assistant · Radiation Oncology Laboratory
  • Expanded a LabVIEW/DAQmx signal chain for P409X respiratory transducers with calibration, voltage-to-pressure conversion, conditioning, digital filtering, real-time graphing, and timestamped recording, improving practitioner breathing-pattern modeling accuracy by 50%.
  • Replaced legacy blocks with current open-source VIs, modularized acquisition and visualization as reusable SubVIs, and increased the amount of respiratory data available for abnormal-waveform analysis.
  • Created MATLAB functions to organize 5DCT patient datasets, improve artifact visibility, flag likely registration errors, and separate false positives for researcher review.
Jun 2024 — Aug 2024

Boeing

Core Analytics & Maintenance Department Intern
  • Reviewed 787 maintenance manuals, service records, and customer-reported issues to classify recurring installation and damaged-part patterns for the Core Analytics team.
  • Refined damaged-part routing and maintenance-record retrieval so technician entries reached the correct engineering review path with more relevant historical context.
  • Validated an internal Generative AI application against known maintenance cases and refined its data pipeline to surface more relevant records, improving processing efficiency by 20%.
03 / TECHNICAL TOOLBOX

Skills grouped by how I use them.

No proficiency bars—just the tools and methods I’ve applied directly in the work above.

Mechanical Design & Manufacturing

Designing mechanisms that can be built, assembled, and tested.

SolidWorksMechanical assembliesMechanism designKinematicsDFMTolerance stack-upMaterial selectionProcess planning

Instrumentation & Analysis

Turning sensor signals and imaging data into usable engineering information.

LabVIEWDAQmxMATLABSensor acquisitionCalibrationSignal conditioningDigital filteringMathematical modeling

Software & Data

Building full-stack applications, APIs, and validation pipelines.

PythonC++JavaReactTypeScriptNode.jsFastifyPostgreSQLAPI designAzure

Product Development

Carrying technical work from an initial need to a verified deliverable.

Requirements gatheringTechnical specificationsDesign documentationVerification planningUser-centered designCross-functional workTechnical presentations

04 / CONTACT

Let’s build something useful.

I’m interested in mechanical design, robotics, medical technology, technical product development, and software that connects to real physical systems.