diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 2c66c49..fe2b422 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -20,7 +20,7 @@ jobs:
       - name: Install build tools
         run: |
           sudo apt-get update
-          sudo apt-get install -y build-essential g++-14 clang clang-format
+          sudo apt-get install -y build-essential g++-14 clang clang-format libtbb-dev
 
       # 3. Clang-format check
       - name: Clang-format Check
diff --git a/README.md b/README.md
index d6e0ec6..ae035fe 100644
--- a/README.md
+++ b/README.md
@@ -35,6 +35,7 @@ Examples include (and will expand to):
     * Encoding
         * [rot47](./rot47/)
     * [prefix-sum](./prefix-sum/)
+    * [sudoku-solver](./sudoku-solver/)
     * [pi-monte-carlo](./pi-monte-carlo/)
     * [pi](./pi)
 * Data Structures
diff --git a/parallel-transform/Makefile b/parallel-transform/Makefile
index 172da2f..e51a197 100644
--- a/parallel-transform/Makefile
+++ b/parallel-transform/Makefile
@@ -3,6 +3,7 @@ include ../common.mk
 
 # per-example flags
 # CXXFLAGS += -pthread
+LDLIBS += -ltbb
 
 ## get it from the folder name
 TARGET := $(notdir $(CURDIR))
@@ -12,7 +13,7 @@ OBJS   := $(SRCS:.cpp=.o)
 all: $(TARGET)
 
 $(TARGET): $(OBJS)
-	$(CXX) $(CXXFLAGS) -o $@ $^
+	$(CXX) $(CXXFLAGS) -o $@ $^ $(LDFLAGS) $(LDLIBS)
 
 %.o: %.cpp
 	$(CXX) $(CXXFLAGS) -c $< -o $@
diff --git a/sudoku-solver/Makefile b/sudoku-solver/Makefile
new file mode 100644
index 0000000..ab8b3c0
--- /dev/null
+++ b/sudoku-solver/Makefile
@@ -0,0 +1,31 @@
+# pull in shared compiler settings
+include ../common.mk
+
+# per-example flags
+# CXXFLAGS += -pthread
+
+## get it from the folder name
+TARGET := $(notdir $(CURDIR))
+## all *.cpp files in this folder
+SRCS   := $(wildcard *.cpp)
+OBJS   := $(SRCS:.cpp=.o)
+
+all: $(TARGET)
+
+$(TARGET): $(OBJS)
+	$(CXX) $(CXXFLAGS) -o $@ $^
+
+%.o: %.cpp
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
+run: $(TARGET)
+	./$(TARGET) $(ARGS)
+
+clean:
+	rm -f $(OBJS) $(TARGET)
+
+# Delegates to top-level Makefile
+check-format:
+	$(MAKE) -f ../Makefile check-format DIR=$(CURDIR)
+
+.PHONY: all clean run check-format
diff --git a/sudoku-solver/main.cpp b/sudoku-solver/main.cpp
new file mode 100644
index 0000000..c0c9877
--- /dev/null
+++ b/sudoku-solver/main.cpp
@@ -0,0 +1,272 @@
+/**
+DFS Sudoku solver CLI.
+
+Input:
+  argv[1]: 81-char board string. Use digits 1-9 for filled cells and '.' or '0' for empty.
+  argv[2] (optional): max number of solutions to print (positive integer).
+
+Output:
+  Prints each full solution as a single 81-char line.
+*/
+
+#include <array>
+#include <cstdint>
+#include <iostream>
+#include <optional>
+#include <string>
+#include <string_view>
+#include <vector>
+
+namespace {
+
+struct SudokuState
+{
+    std::array<int, 81> cells{};
+    std::array<uint16_t, 9> row_mask{};
+    std::array<uint16_t, 9> col_mask{};
+    std::array<uint16_t, 9> box_mask{};
+};
+
+constexpr uint16_t kAllDigitsMask = 0x03FE; // bits 1..9
+
+int
+box_index(int row, int col)
+{
+    return (row / 3) * 3 + (col / 3);
+}
+
+bool
+place_digit(SudokuState& state, int index, int digit)
+{
+    const int row = index / 9;
+    const int col = index % 9;
+    const int box = box_index(row, col);
+    const uint16_t bit = static_cast<uint16_t>(1u << digit);
+
+    if ((state.row_mask[row] & bit) != 0 || (state.col_mask[col] & bit) != 0 || (state.box_mask[box] & bit) != 0) {
+        return false;
+    }
+
+    state.cells[index] = digit;
+    state.row_mask[row] |= bit;
+    state.col_mask[col] |= bit;
+    state.box_mask[box] |= bit;
+    return true;
+}
+
+void
+remove_digit(SudokuState& state, int index, int digit)
+{
+    const int row = index / 9;
+    const int col = index % 9;
+    const int box = box_index(row, col);
+    const uint16_t bit = static_cast<uint16_t>(1u << digit);
+
+    state.cells[index] = 0;
+    state.row_mask[row] &= static_cast<uint16_t>(~bit);
+    state.col_mask[col] &= static_cast<uint16_t>(~bit);
+    state.box_mask[box] &= static_cast<uint16_t>(~bit);
+}
+
+uint16_t
+candidate_mask(const SudokuState& state, int index)
+{
+    const int row = index / 9;
+    const int col = index % 9;
+    const int box = box_index(row, col);
+    const uint16_t used = static_cast<uint16_t>(state.row_mask[row] | state.col_mask[col] | state.box_mask[box]);
+    return static_cast<uint16_t>(kAllDigitsMask & static_cast<uint16_t>(~used));
+}
+
+int
+popcount16(uint16_t x)
+{
+    int count = 0;
+    while (x != 0) {
+        x = static_cast<uint16_t>(x & static_cast<uint16_t>(x - 1));
+        ++count;
+    }
+    return count;
+}
+
+std::optional<SudokuState>
+parse_board(std::string_view input)
+{
+    if (input.size() != 81) {
+        return std::nullopt;
+    }
+
+    SudokuState state{};
+    for (size_t i = 0; i < input.size(); ++i) {
+        const char c = input[i];
+        if (c == '.' || c == '0') {
+            state.cells[i] = 0;
+            continue;
+        }
+        if (c < '1' || c > '9') {
+            return std::nullopt;
+        }
+
+        const int digit = c - '0';
+        if (!place_digit(state, static_cast<int>(i), digit)) {
+            return std::nullopt;
+        }
+    }
+
+    return state;
+}
+
+std::string
+board_to_string(const SudokuState& state)
+{
+    std::string out;
+    out.reserve(81);
+    for (int value : state.cells) {
+        out.push_back(static_cast<char>('0' + value));
+    }
+    return out;
+}
+
+bool
+choose_next_cell(const SudokuState& state, int& out_index)
+{
+    int best_index = -1;
+    int best_count = 10;
+
+    for (int i = 0; i < 81; ++i) {
+        if (state.cells[i] != 0) {
+            continue;
+        }
+
+        const uint16_t mask = candidate_mask(state, i);
+        const int count = popcount16(mask);
+
+        if (count == 0) {
+            out_index = -1;
+            return true;
+        }
+        if (count < best_count) {
+            best_count = count;
+            best_index = i;
+            if (best_count == 1) {
+                break;
+            }
+        }
+    }
+
+    out_index = best_index;
+    return false;
+}
+
+void
+solve_dfs(SudokuState& state, std::vector<std::string>& solutions, size_t max_solutions)
+{
+    if (max_solutions != 0 && solutions.size() >= max_solutions) {
+        return;
+    }
+
+    int index = -1;
+    const bool dead_end = choose_next_cell(state, index);
+
+    if (dead_end) {
+        return;
+    }
+
+    if (index == -1) {
+        solutions.push_back(board_to_string(state));
+        return;
+    }
+
+    uint16_t mask = candidate_mask(state, index);
+    while (mask != 0) {
+        const uint16_t bit = static_cast<uint16_t>(mask & static_cast<uint16_t>(-static_cast<int16_t>(mask)));
+
+        int digit = 1;
+        while ((bit & static_cast<uint16_t>(1u << digit)) == 0) {
+            ++digit;
+        }
+
+        if (place_digit(state, index, digit)) {
+            solve_dfs(state, solutions, max_solutions);
+            remove_digit(state, index, digit);
+
+            if (max_solutions != 0 && solutions.size() >= max_solutions) {
+                return;
+            }
+        }
+
+        mask = static_cast<uint16_t>(mask & static_cast<uint16_t>(mask - 1));
+    }
+}
+
+std::optional<size_t>
+parse_positive_limit(const std::string& s)
+{
+    if (s.empty()) {
+        return std::nullopt;
+    }
+
+    size_t value = 0;
+    for (char c : s) {
+        if (c < '0' || c > '9') {
+            return std::nullopt;
+        }
+        value = value * 10 + static_cast<size_t>(c - '0');
+    }
+
+    if (value == 0) {
+        return std::nullopt;
+    }
+
+    return value;
+}
+
+void
+print_usage(const char* program)
+{
+    std::cerr << "Usage: " << program << " <81-char-board> [max_solutions]\\n"
+              << "Board chars: 1-9 for fixed cells, '.' or '0' for empty cells.\\n";
+}
+
+} // namespace
+
+int
+main(int argc, char* argv[])
+{
+    if (argc < 2 || argc > 3) {
+        print_usage(argv[0]);
+        return 0;
+    }
+
+    size_t max_solutions = 0; // 0 means unlimited
+    if (argc == 3) {
+        const auto parsed_limit = parse_positive_limit(argv[2]);
+        if (!parsed_limit.has_value()) {
+            std::cerr << "Error: max_solutions must be a positive integer.\\n";
+            return 1;
+        }
+        max_solutions = *parsed_limit;
+    }
+
+    auto parsed_state = parse_board(argv[1]);
+    if (!parsed_state.has_value()) {
+        std::cerr << "Error: invalid board input. Expected 81 chars and no row/column/box conflicts.\\n";
+        return 1;
+    }
+
+    SudokuState state = *parsed_state;
+    std::vector<std::string> solutions;
+    solve_dfs(state, solutions, max_solutions);
+
+    if (solutions.empty()) {
+        std::cout << "No solutions found.\\n";
+        return 0;
+    }
+
+    std::cout << "Found " << solutions.size() << " solution(s).\\n";
+    for (const auto& solution : solutions) {
+        std::cout << solution << "\\n";
+    }
+
+    return 0;
+}
diff --git a/sudoku-solver/tests.sh b/sudoku-solver/tests.sh
new file mode 100755
index 0000000..fefa6e9
--- /dev/null
+++ b/sudoku-solver/tests.sh
@@ -0,0 +1,26 @@
+#!/bin/bash
+
+set -ex
+
+puzzle="53..7....6..195....98....6.8...6...34..8.3..17...2...6.6....28....419..5....8..79"
+expected="534678912672195348198342567859761423426853791713924856961537284287419635345286179"
+
+output=$(./sudoku-solver "$puzzle")
+if ! echo "$output" | grep -Fq "$expected"; then
+    echo "Test failed: expected solution was not found"
+    exit 1
+fi
+
+invalid="11..............................................................................."
+if ./sudoku-solver "$invalid" >/dev/null 2>&1; then
+    echo "Test failed: contradictory board should be rejected as invalid input"
+    exit 1
+fi
+
+limited_output=$(./sudoku-solver "$puzzle" 1)
+if ! echo "$limited_output" | grep -Fq "Found 1 solution(s)."; then
+    echo "Test failed: max_solutions limit did not work"
+    exit 1
+fi
+
+echo "All tests passed"