Customer meetings / solution introduction
Listen to AI adoption goals and security concerns, then explain where Sapie Guardian fits and how it can be applied.
English Overview
Jaehwan Ha | Solutions Engineer / Technical BD
I connect GenAI security solutions to customer problems, technical validation, proposal messaging, pricing strategy, and deal-decision evidence.
Current Work
Sapie Guardian customer-facing SE / Technical BD work comes first.
Current work centers on turning GenAI security concerns into customer meetings, technical answers, proposal materials, pricing logic, and trust evidence.
Technical Q&A / fit judgment
Answer practical questions about masking, LLM guardrails, APIs, integration boundaries, and operating constraints.
Sales kits / proposal messaging
Separate public-sector and private-sector materials, then refine proposal messages from customer feedback.
Pricing strategy / quotation support
Shape quotation logic around customer budget, product positioning, and government-program fit.
Trust documents / certification support
Support GS certification materials and trust documents that customers can use during review.
Inbound AI SI technical review
Review inbound AI SI opportunities for technical feasibility, solution fit, and early Q&A readiness.
Current Deliverables
Customer-ready materials and decision evidence.
01Customer meeting notes / requirements summaries
02Solution introduction decks
03Technical Q&A notes
04Public / private sales kits
05Proposal messages / pitch slides
06Quotation and pricing strategy memos
07GS certification and trust documents
08Inbound AI SI technical review notes
Selected Cases
Representative cases that support the current SE / BD role.
AICC Chatbot → Callbot PoC
Turned customer requirements into PoC scope, API behavior, response-time validation, and cost-risk judgment.
Customer issue: response time, multi-turn behavior, and operating cost risk.SE/BD role: requirements, PoC scope, RAG chatbot, callbot-compatible API, and pricing-risk review.Evidence: about 250 official support docs/FAQ items; per validation materials delivered to the customer, 96.7% single-turn accuracy, 95% multi-turn accuracy, and cache hits around 0.1s.Decision point: identified AWS/LLM usage-fee variability risk and proposed a pricing structure that lowers per-channel cost.Smart-Home Wall-Pad Voice AI — Validation & Scope Control → Win
Separated what is technically possible from what this contract can deliver, to win and safely execute the deal.
Customer issue: embedded wall-pad voice AI required a clear line between technical feasibility and contract-deliverable scope.SE/BD role: benchmarked Korean TTS candidates (gTTS, MeloTTS, Zonos, CosyVoice, Kokoro, Coqui) on language quality, install complexity, runtime, and latency, then realigned scope against meeting notes, feature specs, and the contract.Evidence: TTS benchmark comparison, reduced PoC scope (core device control + sLLM/RAG validation), best-effort latency targets, post-support boundaries (defects vs. new requests).Decision point: won and delivered — controlled schedule, scope-expansion, and free-support risks at the contract stage.idus Hyper-Personalization → Sapie-Reco
Connected a customer recommendation problem to a productizable recommendation flow.
Customer issue: gift recommendation for cold users with relationship, occasion, budget, and preference context.SE/BD role: requirements, scope, prototype planning, and productization handoff as PM.Evidence: conversational condition extraction and metadata-based recommendation filtering.Decision point: led to Sapie-Reco productization, commercial rollout, and AWS Marketplace global listing; Sapie-Reco was selected as an outstanding solution, earning booth sponsorship for AWS AI x Industry Week 2025.NIPA Semiconductor Manufacturing AI Agent Program
Connected technical composition, proposal message, architecture, and budget structure into a Technical BD case.
Customer issue: policy fit, industrial impact, execution plan, and budget logic had to work together.SE/BD role: proposal direction, technical architecture, pitch materials, and pre-sales messaging as a team contributor.Evidence: four workflows, 21 specialized AI agents, and about KRW 3.1B cumulative scope.Decision point: passed roughly 20:1 written screening and was selected in the 3:1 final pitch as a team proposal contribution.Experience Summary
Condensed around current customer-facing work.
2026.01-Present
GenAI Solutions Engineer / Technical BD
Customer-facing Sapie Guardian SE / Technical BD work (customer meetings, technical Q&A, sales kits, pricing strategy, B2G requirement review, proposals), plus a solo enterprise callbot PoC (sales, build, validation, pricing), smart-home (wall-pad) voice AI deal support, and NIPA semiconductor AI Agent proposal contribution.
2024.02-2025.12
AI Service Development
Built and validated RAG/LLM chatbots, recommendation flows, and AI Agent PoCs; contributed to Sapie-Reco and Sapie-Braille productization and recognition evidence.
2023.02-2024.01
App / IoT Development
Built Flutter, Firebase, and Raspberry Pi based smart-farm app and IoT structures.
Resume Documents
Purpose-specific resume files.
Proof / Awards
Five public proof signals.
01
Grand Prize, AI Model Development, Korea Fair Trade Commission
02
NIPA AI Agent program selection contribution
03
Sapie-Braille Public Procurement Service encouragement award
04
Sapie-Reco selected as an outstanding solution — booth sponsorship for AWS AI x Industry Week 2025, AWS Marketplace global listing
05
AWS and Databricks certifications
Public Technical Proof
Sanitized technical evidence.
Customer-private PoCs are not exposed; public proof is limited to sanitized RAG, document automation, and AI-assisted workflow evidence.
01
Sapie-Braille accessibility AI PM
02Public proof around an accessibility AI that combines standards-based Braille conversion with AI (document understanding, alt text, voice interface, information-navigation agents).
AI sales follow-up agent
03An Excel/CSV-first contact workflow (column mapping, row-level review) with pre-send checks and Outlook draft generation (draft-only by default); multimodal LLM business-card extraction is kept as a secondary path after early hypothesis validation.
Admissions RAG chatbot backend
04Publicly shareable RAG backend evidence around document ingestion, retrieval, and response boundaries.
Public GitHub proof package
A safe proof package that separates customer-private PoCs from public RAG, document automation, and AI-assisted development evidence.
Contact