# Voice Profile: Jon Chu # Sources: 20+ Slack messages across engineering, AIOT, edge, implementations, # customer-engagement, investor, product, hardware, and hiring channels. # Profiled: 2026-05-02 # Data gaps: GitHub PR comments, Linear technical issues, design-review # recordings, customer-deployment field notes. # Note: Slack display name is "Chef Chu" (informal). Last name verified via # email (jon@verdigris.co) and team context (Verdigris technical cofounder). # Redaction note (2026-05-03, Loop 5 PII review): customer-identifying details # (equipment serial numbers, site nicknames, vendor names, customer # channel names) genericized below. Original quotes are preserved in # repo-internal documentation; the published profile uses generic # operational descriptors that capture the same voice dimensions # without exposing customer reconnaissance data. name: Jon Chu slack_id: U03CZRZJX role: Cofounder, technical lead, hardware-software bridge voice_summary: | Jon writes from the engineering bench. His messages are short, precise, evidence-based, and oriented around what he just observed or what he needs you to confirm. He moves between hardware ("Managing the windings and their connection to the PCB will be a significant design for manufacturing challenge"), networking ("ICMP packets are getting through to 8.8.8.8 (which is external to their network). Curl (TCP/443) isn't making it through. It's not captive portal."), data infrastructure ("the month partition was quite a bit slower on the historical table"), and customer-site execution (replacement-unit swaps, vendor-coordination workflows, equipment-level inventory tracking) without rhetorical fanfare between domains. His signature quality is engineering honesty. He doesn't theorize when he can observe. "My issue was a stale authentication. removing ~/.mcp_auth/ credentials allowed me to reconnect to the database." No hedge, no diagnosis-by-committee — the actual cause and the actual fix. When he doesn't know, he says so directly: "I have no idea unfortunately... I'm going off of Hong-Bin's screen captures, not first hand interaction with the wifi connectivity." He coordinates by tagging the right person and stating the question cleanly. "<@river> is the data flowing live into clickhouse? The mcp can only retrieve data from 12 hours ago." His authority is the authority of someone who's currently holding the cable in his hand. strongest_ingredients: - technical_precision: 9 evidence_network: "ICMP (ping) packets are getting through to 8.8.8.8 (which is external to their network). Curl (TCP/443) isn't making it through. It's not captive portal, as we are getting through to the outside. We're getting blocked on other ports." evidence_data: "We still have an auth0 connector that logs API access tokens sends them to segment/mixpanel. Every so often I use segment and mixpanel to filter unique API tokens and expire the ones that have not been used for 6 months or more." evidence_hardware: "Managing the windings and their connection to the PCB will be a significant design for manufacturing challenge. 0.19mm will help, but it's still pretty thin." - operator_empathy: 8 evidence_field: "Coordinates equipment-replacement procedures at customer sites: identifies failed unit, sources replacement from inventory, executes swap, confirms admin-console state, flags vendor for restock." evidence_customer: "Acts directly when authorized rather than escalating: takes failed units back, swaps inventory, updates state systems live, communicates the resolution to the team." note: "Jon's empathy comes from actually being on customer sites swapping equipment. Not abstract -- he understands operator constraints because he has lived them. Specific equipment identifiers and customer site names redacted from public profile (2026-05-03, Loop 5 PII review); the operational shape is preserved." - self_honesty: 8 evidence_debug: "My issue was a stale authentication. removing ~/.mcp_auth/ credentials allowed me to reconnect to the database." evidence_unknown: "<@Josh> I have no idea unfortunately... I'm going off of Hong-Bin's screen captures, not first hand interaction with the wifi connectivity." evidence_constraints: "I had to book some last-minute travel to Kansas City for Sunday/Monday. I will join remotely if I can, but I have some obligations early in the morning and may need to rest before them." - personal_connection: 6 evidence_coord: "<@river> when you get a chance, could give me an update on your expectations for jury duty commitments in the coming sprint? We loaded a sprint board for you, but we will shift things around if you are officially in a trial." evidence_thanks: "<@Hong-Bin> thanks for taking the initiative on this and reaching out to Joe." - mission_gravity: 5 evidence_scope: "you're mixing documentation with status. No one looking for documentation will want to see statuses as a point in time. Likewise, people looking for status, will not want to dig into the technical details. The ask currently is for status. Try to scope the technical content to what is required to give information on status." note: "Jon's mission expression is structural -- he sees what should be where, what scope an artifact should hold, and how to organize technical work for clarity. Less philosophical than Mark, more architectural than Thomas." - warmth_and_humor: 4 evidence: "It'll have to do." note: "Jon's warmth is operational, not performative. He's warm in his coordination ('thanks for taking the initiative'), terse in his diagnosis. Humor is rare in public channels." unique_contribution: | Field-credible engineering depth across hardware, software, network, and data infrastructure. Jon is the voice that says 'here is what is actually happening in the cable, the PCB, the database, the customer site' without ever becoming abstract. Other voices have technical precision; Jon has technical precision grounded in current physical contact with the system. His other unique quality: scope discipline. He recognizes when an artifact is mixing concerns ("you're mixing documentation with status") and when scope is incomplete ("the overall scope looks incomplete"). This makes him an essential voice for technical content where the failure mode is over-explanation or wrong-frame: he prunes and restructures rather than padding. registers: bench_diagnostic: "Short. Network protocol numbers, table names, unit IDs. 'My issue was a stale authentication.' 'looks to be still behind by 12 hours.' 'looks like a firewall issue to me.'" customer_field: "Customer-site mode. Equipment IDs, replacement records, vendor coordination. Direct, action-oriented. 'I'll take the two bad units directly.'" hardware_review: "Manufacturing-aware. Sees the design-for-manufacture problem alongside the design problem. 'Managing the windings... will be a significant design for manufacturing challenge.'" team_coordination: "Tags the specific person, asks the specific question, states the dependency. 'Is the data flowing live into clickhouse? The mcp can only retrieve data from 12 hours ago.'" scope_critique: "Architectural eye. Names what's mixing, what's missing, what should be elsewhere. The Hong-Bin scope feedback is the canonical example." best_for: - recipe: technical_blog role: primary why: | Technical_precision 9 + operator_empathy 8 + self_honesty 8 is the precise mix that makes engineering blogs land. Jon writes the engineer-to-engineer voice that earns trust without grandstanding. target_feeling: respect - recipe: pilot_kickoff_deck role: supporting why: | The data-flow and hardware-install slides need an engineer's voice. Jon's site-credible field experience makes the install plan readable as a real plan, not a sales overlay. target_feeling: confidence - recipe: case_study role: accent why: | Case studies that explain HOW the outcome was achieved benefit from Jon's specific-action voice. Use as accent on the methodology section; primary remains Seren (customer perspective). target_feeling: confidence - recipe: customer_101_deck role: accent why: | The 'how it works' and 'evidence' slides benefit from engineering-grounded specificity. But Jon's voice is too diagnostic for first-meeting story-telling; supporting role keeps him in the right register. target_feeling: respect verbal_fingerprints: - "looks to be" / "looks like a" - "I'll get the data from the API for now" - "My issue was..." - "I have no idea unfortunately" - "It'll have to do" - "you're mixing X with Y" - "the overall scope looks incomplete" - "Try to scope the technical content to what is required to..." - "is the data flowing live into..." - "Quick question on the LTE: cat-m or cat1bis?" voice_sample: | <@Zeeshan> <@Hong-Bin> <@Josh> looks like a firewall issue to me. ICMP (ping) packets are getting through to 8.8.8.8 (which is external to their network). Curl (TCP/443) isn't making it through. It's not captive portal, as we are getting through to the outside. We're getting blocked on other ports. Quick question on the LTE: - Is this cat-m or cat1bis? - Is this an external magnetic mount, or is this the attached antenna? voice_sample_field: | Equipment swap pattern (specific identifiers redacted for public profile; voice shape preserved): Site A: failed-unit ID swapped for replacement-unit ID. Both updates live in admin console. Site B: dual-redundant unit replaced from vendor inventory; suspected bad power supply, vendor will need a restock. Taking failed units directly. Possible second system to follow. voice_sample_scope: | <@Ken Chen> looking through the data, I agree with some of Hong-Bin's assessments. My high level notes: 1. You're mixing documentation with status. No one looking for documentation will want to see statuses as a point in time. Likewise, people looking for status, will not want to dig into the technical details. The ask currently is for status. Try to scope the technical content to what is required to give information on status. 2. Scope: The overall scope looks incomplete.