I designed and built Overview end to end, bringing crop health, imagery history, growth stages, and weather into a platform that helps people find the information behind a field decision.
Founder · sole designer and developer of the platform
System
Satellite indices → crop analysis → alerts and field actions
Surfaces
Web · iOS · Android
The challenge
As imagery, crop analysis, weather, and field tools came together, the hardest problem was helping people stay oriented. A broad feature set needed to feel approachable, with a clear path to the information each user was looking for.
The design decision
I built a question-based navigation approach. A series of questions directs users to the relevant workspace or information, helping them reach an answer without first learning the application's navigation structure.
Early testing
Gradient Crop Yield Solutions was an early tester of the platform.
Inside the system +
Overview began as Honeycomb Robotics' crop-monitoring application. Its grower and technician workspaces now bring together five spectral views—NDVI, NDRE, MSAVI, NDMI, and RGB—with over eight years of imagery history.
Automated stress alerts and field status labels.
A growth-stage model combining growing degree days with satellite indices.
Pro tools for scouting tasks, optimized routes, prescription maps, application records, and field intelligence chat.
English and Spanish mobile apps. Tomato yield and growth-stage models are available; almond models are coming soon.
The operation at a glance / All Fields DashboardField intelligence from the truck. iOS & Android · English & Spanish
I solely built the web, iOS, Android, and backend applications, tailoring each to its platform and use case. Working with more than 20 mechanics and group leaders, I developed a system for tracking the parts replaced in the field.
REQUIREMENTSApril 2026
→
IN PRODUCTION2026 harvest
Scan
Read a part’s QR code or barcode in the field.
Assign
Connect that part to a harvester, tractor, or other machine.
Trace
Preserve the assignment audit trail, parts cost, and repair history.
Report
Track mechanic progress and equipment spending week over week.
My role
Sole developer · requirements, web, backend, iOS, Android, and rollout
System
Field replacements → trailer stock and machine history → weekly operations reporting
Surfaces
Web dashboard · Swift iOS app · React Native / Expo Android app
Field workflow
Recording a replacement was simplified to a QR-scanning process that can be completed in under a minute. Field teams can connect a replacement to the machine and mechanic, while offline scanning supports work in low-signal areas.
Parts visibility
Parts administrators can see which field trailers are low on stock or out of stock. Replacement records show which parts are replaced most often, who performed the work, and the expenditures associated with machine breakdowns.
Weekly reporting
Reports give administration and group leaders a shared view of replacement activity and spending, including which parts were replaced, which mechanics performed the work, and the associated expenditures.
Web
TypeScript · Next.js · Tailwind CSS · shadcn/ui
API
Python · FastAPI
iOS
Swift
Android
React Native · Expo
Hosting
AWS
Inside the delivery +
Built at Honeycomb Robotics for Cal Sun Grower Services, part of The Morning Star Company. Delivery included dry runs, colleague training, and in-field support.
Offline QR and barcode scanning to assign parts to equipment.
Role-based access and a complete assignment audit trail.
Low-stock alerts, reorder points, and machine-level parts and repair costs.
Weekly mechanic progress and equipment expenditure reporting.
Next phase: office QR scanning, direct T-STAR integration, vendor and order management, and in-app reordering.
PROJECT / YIELD PREDICTIONIn validation
Forecast the yield. Show the uncertainty.
A regression-based system for estimating processing-tomato yield from Sentinel-2 indices, weather, and transplant dates—with an 80% prediction range. The model is still in validation.
AVERAGE ERROR AT DAY 80 TONS / ACRE · LOWER IS BETTER
County baseline12.2 t/ac
Yield model9.1 t/ac
26% lower average error than the county baseline at day 80. Improvement in every season tested.
My role
Model development and season-by-season validation, with Robby Force at The Morning Star Company
System
Cloud-masked satellite indices + weather + transplant date → field-level yield estimate
Surfaces
5,188 field-seasons · 2021–2024 · October 2026 report
Data at scale
I worked with Morning Star field data across multiple years, processing hundreds of millions of data points spanning weather, crop health, and yield information to build the prediction system.
The engineering challenge
The hardest part was working with the volume of data: understanding how to query it and building a system that supplies the model with the information it needs, when it needs it.
Intended use
The system is intended to help business-development and factory teams understand yield potential at the field level and bring those estimates into their planning.
Validation and next steps +
Each season was predicted by a model trained on the other three seasons, using only information available at the prediction date. At day 80, 66% of fields were within ±10 t/ac and 81% of actual yields fell inside the model's 80% prediction range.
Season-by-season holdout results
Checkpoint
Error (t/ac)
Within ±10 t/ac
Inside 80% range
Day 60
9.8
61%
82%
Day 80
9.1
66%
81%
Day 100
8.9
67%
82%
Harvest
8.8
67%
81%
County baseline
12.2
50%
—
Next: add the 2025 season, stable field IDs, variety, and harvest dates; forward-test on 2026 before running in-season in 2027.
Build the system
Design the architecture, write the code, and ship it.
Own the delivery
Take requirements through release, training, and field support.
Measure the result
Use validation, operating outcomes, and economic feasibility.
MORE SYSTEMS
More work. Different operational problems.
Agriculture, logistics, and device control—from production software to prototypes and feasibility studies.
One surface to plan, execute, and reconcile the season. The platform combines 10-second GPS tracking, equipment and crew records, OSRM field sequencing, schedule-change approvals, and season-over-season yield tracking.
Deterministic simulation of field paths, split crews, truck cycles, FIFO factory unloading, and variety-aware processing windows.
Animated map playback from 1× to 500× with a crew Gantt and factory capacity gauges.
Regular, overtime, and double-time labor-cost analysis under California Wage Order 14.
Irrigation block editing and ETo-driven schedules; soil-moisture, weather, and pressure-event monitoring with SSURGO soil inference.
Cal Sun personnel managementHarvest staffing / Honeycomb RoboticsIn development+
Mockups complete; development underway. A harvest staffing platform replacing calls, texts, and spreadsheets with a schedule board for each machine's crew, shift, and field.
Replacement recommendations based on qualifications, distance, and rest limits.
Relocation cost comparisons covering mileage, per diem, and lodging, routed to logistics for approval.
Hotel management and a Spanish-language operator app with one-tap shift confirmation.
Weekly group cost reporting and a scenario builder.
Built a custom warehouse management and fulfillment application at Honeycomb Robotics. A rate-shopping engine aggregates live UPS, USPS, and DHL rates and selects the lowest-cost option for each shipment.
Automated medication dispensingDevice control / Confidential clientPilot+
Developed control software that enforces prescription-based dispensing limits and detects and reports tampering. In a one-patient New York pilot, the physician could review medication times and device events.
The software performed as designed; the pilot identified mechanical revisions needed in the device.
Drone foreign-object detectionComputer vision / Confidential herb processorFeasibility study+
Surveyed fields and trained a model on thousands of drone images, reaching approximately 67% detection accuracy. The economic study found drone detection with manual cleanup crews was not feasible.
Recommended redirecting the solution toward foreign-object ejection during harvest.
Stomato field service platformThe Morning Star Company / 2020–2024Field systems+
Built database storage, automated data intake and page setup, multi-crop support, a landing page, and field-technician alerts.
Serviced acreage grew from approximately 11,600 to 26,000 without expanding the software team.
Led development of the full-stack platform and service website, saving an estimated $120,000–$150,000 by building in-house. Worked with stakeholders and the company owner on operations, economics, expansion strategy, annual roadmaps, and reviews.
Aerial imaging with UC MercedMorning Star R&D / University collaborationProof of concept+
Guided four senior computer science students in building a web application to create and view NDVI crop-health images from satellite data. Delivered a proof of concept valued at approximately $30,000.
EXPERIENCE / ENGINEERING + LEADERSHIP
Engineering, with operational ownership.
Building the software is one part of the work. I also lead technology direction, plan releases, work with stakeholders, and support adoption in the field.
Building satellite crop intelligence with Overview and developing Control for harvest planning, simulation, and logistics.
FEB 2024 — PRESENT
Chief Technology Officer
Honeycomb Robotics
Direct infrastructure, software development, and cybersecurity across company products and client projects. Deliver field systems from requirements through release, training, and operational support.
JAN 2020 — FEB 2024
Co-founder
Gradient Crop Yield Solutions · The Morning Star Company
Led full-stack platform development and annual feature roadmaps, working with stakeholders on operations, economics, and expansion strategy.
JAN 2020 — JAN 2024
Software / R&D Engineer
The Morning Star Company
Built Stomato’s data storage, automated intake, multi-crop support, and technician alerts. Mentored a four-student UC Merced satellite-imaging team.
Overall tenure at The Morning Star Company: January 2018–February 2024.
COMPANY / FOUNDER · 2026–PRESENT
Sentilli Analytics
I founded Sentilli Analytics to bring crop monitoring, field intelligence, and operational planning into one product family for growers and processors.
California's Central Valley Engineer · founder · technology leader
Understand the operation. Then build the system.
I bring eight years of experience connecting software engineering with the practical demands of agricultural operations. My work spans full-stack web and native mobile apps, geospatial systems, cloud infrastructure, and applied machine learning.
I work across the delivery cycle: understanding the problem, designing and building the system, and supporting the people who use it. Beyond agriculture, I've delivered client software in logistics and medical devices.
UC MERCED / MAY 2018
B.S. Computer Science and Engineering Minor in Management and Business Economics
