Manually exporting and reviewing content performance was easy to forget and gave only a single snapshot in time, with no historical growth tracking per post.

Built a content analytics pipeline from scratch: a parser that turns raw export files into structured, deduplicated data (master dataset, daily totals, per-post growth history), rebuilt mid-project with defensive multi-format parsing after the source platform changed its export format without warning. Shipped a full working dashboard (KPI cards, engagement by category, account-wide timeline, per-post growth pages, date-range filtering). Took it from private tool to public product: landing page, About/roadmap, working signup flow. Chose Cloudflare Workers for the backend to consolidate onto existing infrastructure and documented that tradeoff.

Turned a manual, easy-to-forget export process into a real historical record tracking per-post growth over time. Delivered a concrete, demonstrable working tool, not just a resume line. Positioned for both portfolio showcase and live beta-tester recruitment.

Solo build: product, architecture, and public launch end-to-end.

A 4-vendor Digital Front Door engagement for Ascension Health was at risk, with reactive escalation and no clear ownership layers. Later, during SIT 3, Oracle HCM APIs were severely delayed (16 of 52 endpoints validated) and stakeholders conflicted over which features — HR Payroll versus Manager Overtime — to prioritize.

Founded Straventis Advisory Group and took program leadership for the engagement. Standardized Agile delivery, backlog planning, and value-based prioritization across four vendors. Built a 4-layer governance operating model (executive, program, project, vendor). Instituted Jira as the single source of truth to eliminate execution swirl. Negotiated a UI workaround for the delayed system and prioritized the highest user-impact features first.

Shifted an at-risk program to a 6-week cadence, delivering 95% of planned content and restoring on-track status. Extended the engagement into a $3M contract. Stabilized the SIT 3 window by driving completion of the 36 delayed endpoints and forming self-organizing Agile squads.

4-vendor delivery, 140K associate end-user base, executive/program/project/vendor governance layers.

A monolithic, rule-based metadata pipeline couldn't keep pace with modern OTT/VOD volume. A later ML matcher migration created a data-integrity deadlock — parent and child series split across catalogs, causing 20x data bloat — and pipeline cost and latency crept up silently despite no job failures.

Replaced the legacy pipeline with an ML-driven, service-oriented architecture on AWS, Snowflake, and Databricks, implementing a Lambda-to-Snowflake SQL API handoff. Halted a migration to prevent database corruption, authorized a tactical data-mirroring fix, then redesigned the ML-to-knowledge-graph data contract. Decoupled the One Ad Platform via a CAM microservice to bypass legacy blockers. Traced a row-explosion and cardinality issue through Databricks and enforced cardinality governance.

Scaled to 240M records/month across 100+ partners and 5M+ MAUs. $7.9M modeled net benefit, $11M legacy OpEx eliminated over four years. Processing cut from 25+ hours to ~30 minutes. Rescued the VOD roadmap, maintained 95% on-time delivery, and prevented a projected $0.5M delay. Protected $500K in at-risk revenue on the OAP launch. Stabilized pipeline latency back under 30 minutes after cost drifted 18%.

Owned platform architecture and delivery across 4 engineering teams.

Certification and royalty tracking was manual — PDFs, emails, SharePoint — with no lab-partner systemization, causing roughly $800K–$1M/year in missed SKU/royalty revenue, a 70% manual-flow error rate, and a ~45-minute admin burden per request. Separately, AWS usage had no centralized governance, manual account provisioning took days to weeks, and a RAM misconfiguration caused a multi-hour non-production outage.

Owned the Salesforce Global Certification Hub — Internal Console, Experience Cloud Customer Portal, Anonymized Lab Portal, CPQ/Billing — including an anonymized lab-routing engine and zero-touch SDK provisioning via CI/CD. Led the AWS Control Tower implementation, migrating ~90 accounts into a governed OU structure with Terraform-based automated provisioning (Account Factory for Terraform), integrated with ServiceNow, acting as the bridge between Xperi, implementation partner Caylent, and AWS.

Q2C: recovered roughly $800K–$1M/year in revenue, cut processing time 50%, workflow errors 70%, and OpEx 25% (~$500K/year); hit 92% first-time-right SDK delivery. Control Tower: replaced manual provisioning with centralized governance, moved change-request execution from sequential to parallel, and closed the RAM/IPAM outage risk.

Bridge role across Engineering, Sales, IT, Security, FinOps, external lab partners, and AWS/Caylent.

The Ford program needed a clean wrap-up while Harman needed to stand up an entirely new ADAS platform effort from a standing start, with no proven architecture and a fragmented three-party (OEM/Tier 1/Tier 2) supplier ecosystem.

Closed the Ford program (handoffs, wrap-up) while kicking off lidar/computer-vision proof-of-concept work and a data collection campaign. Led the early transition from classical computer vision to CNN-based object detection. Defined a scalable L2+ to L4 autonomous driving architecture on TI Jacinto processors with AUTOSAR Adaptive middleware, freezing interface control documents across OEM/Tier 1/Tier 2. Integrated HD map matching with GNSS/IMU odometry. Architected object-level sensor fusion to decouple perception modalities.

$224M+ in lifetime sales secured on program close. Established the technical foundation for 70%+ software reuse across autonomy tiers. Proved deep learning outperformed heuristic computer-vision models in complex lighting and weather conditions. Unlocked highway-pilot-class localization accuracy.

Cross-functional: OEM, Tier 1, and Tier 2 suppliers on a three-party ecosystem; internal perception and software teams.

Internal pressure pushed an expensive ($1,000+) lidar sensor investment that wasn't market-ready. A vendor's first ML model also dropped accuracy more than 10% when compiled to fit a 2.6 TOPS compute budget three weeks before an OEM demo, and the ADAS ECU failed thermal and EMC testing during design validation.

Built a TAM/SAM/SOM business case and pivoted the architecture to a modular, camera-first Edge-AI platform on Qualcomm/TI Jacinto SoCs. Ran side-by-side bench comparisons of camera-plus-lidar versus camera-only variants. Diagnosed the ML accuracy drop via profiler data (bandwidth saturation), dropped camera resolution, and recalibrated the model with a mixed day/night dataset. Fixed the thermal/EMC failure via dynamic frame rate scaling, power gating, and spread-spectrum clocking, without a hardware redesign. Completed design and process validation and launched both variants as the first committed production line.

Secured $3.2M in internal venture funding. Proved a camera-only fallback, saving $700–900/vehicle in BOM and hedging lidar supply risk. Recovered ML accuracy to a 0.8% drop and 81ms latency, securing mass-production agreements for six vehicle lines. Avoided a $120K board respin. Delivered 2–2.5x revenue uplift via multi-line reuse at L2+ start of production.

Owned platform architecture end-to-end across hardware, firmware, ML, and OEM demo readiness.

A $36M, 3-year global infotainment program originally scoped for US/Mexico manufacturing was thrown into a major scope change when Ford discontinued the US Focus mid-program, forcing a shift to EU/APAC. Mid-program also brought a late spec change (USB-A to USB-C) and a manufacturing flash-time bottleneck.

Realigned factories, suppliers, and launch plans across Focus, Puma, Fiesta, Focus ST (EU), and Ranger (APAC). Managed Tier 1/Tier 2 supplier relationships and interface control documents. Ran a data-driven impact analysis to negotiate a phased USB-C rollout with OEM leadership. Led root-cause analysis and incremental flashing to fix a 50% flash-time spike. Managed OEM and manufacturing-plant relationships across three factories, applying VA/VE to the bill of materials.

100% on-time launch, zero program-caused delays, $224M+ lifetime sales. 22% BOM cost cut without warranty risk. Flash/test time cut 30%, factory throughput restored within two weeks. USB-C conflict resolved without missing the Job 1 date.

Cross-functional: Ford OEM engineering, Tier 1/2 suppliers, three international factories, EMEA/APAC regions.

A startup environment with high organizational chaos and rapidly shifting responsibilities, including being stranded in Japan with delayed payroll mid-assignment.

Built systems integration and HIL testing protocols for instrument cluster electronics. Developed low-level firmware and embedded control logic for automotive and industrial applications. Migrated legacy control software onto a standardized computing platform, abstracting hardware from application logic. Adapted across roles, from Matlab/Simulink modeling to wiring harnesses to translating Japanese engineering emails.

Reduced hardware integration failures in later deployment stages through early validation. Improved system response times with deterministic execution patterns for safety-critical functions. Cut development cycle times for subsequent product lines via a reusable software foundation.

Individual contributor, cross-cultural team with a Japan-based OEM partner.

Engineering troubleshooting relied on guesswork; software defects were found too late in the engine control development cycle, and physical prototype hardware was a scarce validation bottleneck.

Shifted into Technical Program Management, synchronizing hardware, software, and testing teams for a major product launch using structured Agile practices tailored for cyber-physical systems. Architected early telemetry and data-logging pipelines for field-deployed fleets. Built automated simulation environments to validate control algorithms before hardware availability. Built structured test plans and HIL-based automation workflows to isolate issues earlier.

Delivered the program on time. Cut root-cause analysis time ~40% via telemetry. Reduced defect discovery time 25% and post-deployment defects 15% by shifting validation left.

Cross-functional: hardware, software, and test teams on a major product launch.

Teams lacked validation rigor, ownership, and coordination across distributed locations; engine controller platforms needed faster, more trustworthy regression coverage.

Developed automated HIL validation frameworks running fault injection, regression, and system-level validation ahead of physical hardware. Led a 5-person embedded validation team — technical mentoring, work allocation, release coordination. Served as Scrum Master for a cross-functional engineering team. Introduced Agile coaching and Six Sigma DMAIC/DMEDI initiatives. Collaborated across systems, controls, and calibration engineering on production release readiness.

Cut manual engine testing effort ~40% through automation. Improved defect detection 85%, team efficiency 30%, and software reliability 20%. Improved test strategy consistency and release coordination across the team.

Led a 5-person validation team; coordinated across distributed engineering sites.