1. Project Background and Objectives

On May 24, 2019, the World Bank and Ministry of Economy and Finance of the Republic of Uzbekistan (MEF) agreed on a World Bank loan program called “Institutional Capacity Building Project” (ICBP) for Uzbekistan.

The development objectives of ICBP are to strengthen public financial management (PFM) and enable the market operation of the corporate sector. This project has four components. 1) The first component, Improving PFM, aims to support improvements in areas fundamental to a functional PFM system that promotes the achievement of public policy objectives, improved performance, and institutionalized accountability. 2) The second component, Improving Conditions for Market Operation of the Corporate Sector, aims to support the restructuring and rightsizing of the State-Owned Enterprise (SOE) sector in Uzbekistan. 3) The third component, Improving Support Mechanisms for Reforms, aims to provide support to the newly established Economic Council, key regulatory agencies, and other public sector agencies expected to be identified as critical for implementing market reforms during project implementation. 4) The fourth component, Project Management, aims to provide support to the overall implementation arrangements for the project.

As part of this initiative, the Ministry plans to engage a team of experienced local IT specialists.

Pursuant to Instruction No. 25/3‑285 dated 17 June 2025 issued by the Ministry of Economy and Finance, an AI Technologies Working Group—comprising Ministry staff and external experts—will be established to systematically coordinate and strategically advance all AI‑related initiatives. This Working Group will include four local experts: a Network Administrator, a System Administrator, a DevOps Engineer, and a Software Developer.

The primary mandate of these local experts is to enhance the Ministry’s “Online Advisor” service on its official website by leveraging AI technologies to automatically respond to inquiries from individuals and legal entities concerning the Ministry’s core functions—such as state budget preparation and execution, treasury operations, public–private partnerships, subsidy allocation, and other government services.

The Ministry cannot successfully deliver on these objectives without the engagement of qualified and reputable IT professionals with demonstrated experience in these roles: Network Administrator, System Administrator, DevOps Engineer, and Software Developer.

2. Goals and objectives set for the Ministry:

Identify the Ministry’s core service areas (state budget drafting and execution, treasury operations, public–private partnership programs, subsidy allocation, tax and customs regimes, etc.) and compile a comprehensive FAQ corpus reflecting the most frequent inquiries from individuals and legal entities.

Develop and continually update this FAQ corpus using semantic intent recognition and NLP models (e.g., GPT‑series) to ensure relevance and accuracy.

Design and integrate a chatbot module into the Ministry’s official website “Online Advisor” interface, capable of interpreting natural‑language queries, routing users to the appropriate department or document, or providing an automated response.

Enable multilingual support (Uzbek, Russian, English) and implement a continuous feedback‑loop mechanism to improve understanding and response quality over time.

Establish key performance indicators (average response time, successful automated response rate, user satisfaction score) and integrate real‑time monitoring dashboards to track and optimize the “Online Advisor” service.

Improve accessibility and adoption by optimizing UI/UX for both web and mobile platforms, creating user guides, and conducting targeted outreach to stakeholders.

Conduct regular service audits—reviewing chatbot responses, assessing inquiry complexity, and recommending process optimizations—and pilot proposed improvements to validate their effectiveness.

3. Scope of Services:

The Software Developer will:

1. AI Model Integration & Development

– Integrate and adapt pre‑trained NLP/LLM models (e.g., GPT, BERT) into the chatbot backend, ensuring seamless inference pipelines;

– Develop fine‑tuning scripts and training workflows to customize models on Ministry‑specific datasets.

2. Backend API & Microservice Implementation

– Design, implement, and maintain RESTful APIs (FastAPI, Django REST Framework) for chatbot interactions, data retrieval, and reporting;

– Implement authentication, rate‑limiting, and input validation to ensure API security and reliability.

3. Data Processing & Storage

– Build ETL pipelines to ingest, preprocess, and store structured and unstructured data in PostgreSQL and/or MongoDB;

– Ensure data schemas are optimized for low‑latency retrieval and high concurrency.

4. Frontend & Conversational UI Components

– Collaborate with UI/UX teams to develop web and mobile chat interfaces using React, Vue, or similar frameworks;

– Implement real‑time messaging, context management, and multi‑language support (Uzbek, Russian, English).

5. Testing & Quality Assurance

– Write unit, integration, and end‑to‑end tests (pytest, Jest, Selenium) to validate functionality, performance, and security;

– Develop automated test suites for continuous integration pipelines.

4. Qualification requirements

Software Developer to support AI-based digital infrastructure at the Ministry of Economy and Finance.

Requirements for work experience and qualifications:

– Bachelor’s degree in IT/Software Engineering;

– 5+ years of Software experience;

– Strong proficiency in Python;

– Experience with machine learning libraries: Transformers, PyTorch, TensorFlow;

– Ability to develop and integrate RESTful APIs and work with JSON, OAuth2;

– Develop APIs for chatbot interaction, data retrieval, and PDF parsing;

– Integrate Retrieval-Augmented Generation (RAG) models;

– Ensure secure data handling and communication (JWT, TLS);

– Collaborate with frontend and AI teams for seamless delivery.

Candidates possessing the following certifications will be given priority consideration:

AI/ML specialization from recognized platforms (e.g., Google TensorFlow Certification, Coursera, edX);

PCAP – Certified Associate in Python Programming (or equivalent).

5. Responsibility and Performance Assessment

Coordination:

– Reports to the Ministry’s Chief IT Architect and liaises with the Project Implementation Unit (PIU);

– Works on‑site in Tashkent under a jointly‑approved work plan.

Deliverables & Reporting:

– Monthly progress reports detailing network builds, tests, incidents and mitigations;

– Final “as‑built” network diagrams, configuration files and transition documentation.

6. Duration and mode of operation

The duration of the expert's work is expected to be 5 months.

At the same time, the organization of the work of specialists is carried out based on the work plan between The Client and The Expert: the expert works directly in Tashkent. The remuneration for the activities described in these TORs will be in a monthly salary basis.

7. Results

1. Result 1: High-performance GPU servers are procured, configured, and deployed to support AI model training and inference. The AI computing environment is made fully operational for chatbot model execution.

2. Result 2: AI models (e.g., GPT-based) are fine-tuned using ministry-specific content such as legal documents, regulations, and FAQs. The chatbot is trained to support multilingual interaction, including Uzbek.

3. Result 3: The chatbot is connected to core systems such as UzASBO, budget management platforms, and electronic document circulation via secure APIs. Enables real-time responses to data-driven queries (e.g., budget status, payment records, document tracking).

4. Result 4: A responsive web-based interface for the chatbot is developed for ministry staff and the public. Mobile applications for Android/iOS platforms are deployed and user-tested.

5. Result 5: Chatbot queries are protected with authentication, role-based response control, and logging mechanisms. IDS/IPS systems are used to secure AI server network communications.

6. Result 6: Chatbot usage is monitored, and accuracy and user satisfaction levels are measured. Based on feedback and performance reports, the model is retrained, and the response base is regularly updated.

8. Organizational issues

The Client is the Ministry of Economy and Finance. The Ministry is responsible for receiving the work performed by expert within the framework of this assignment, as well as signing the act of acceptance of the work.

The PIU under the Ministry of Economy and Finance is responsible for making payments based on the act of acceptance of work/ Timesheet/Service signed by the Ministry (based on the Protocol approving the relevant deliverable(s)) and the preliminary act of expert.

All costs, including taxes, are covered by expert.